torchhd.datasets

The Torchhd library provides many popular built-in datasets to work with.

class torchhd.datasets.Abalone(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Abalone dataset.

Instances	Attributes	Task	Area
4177	8	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.AcuteInflammation(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Acute Inflammation of urinary bladder dataset.

Instances	Attributes	Task	Area
120	6	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.AcuteNephritis(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Acute Nephritis of renal pelvis origin dataset.

Instances	Attributes	Task	Area
120	6	Classification	Social

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Adult(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Adult dataset.

Instances	Attributes	Task	Area
48842	14	Classification	Social

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.AirfoilSelfNoise(root: str, download: bool = False, transform: Callable | None = None, target_transform: Callable | None = None)

NASA Airfoil Self-Noise dataset. Dataset is obtained from a series of aerodynamic and acoustic tests of two and three-dimensional airfoil blade sections conducted in an anechoic wind tunnel.

Instances	Attributes	Task	Area
1503	6	Regression	Physical

Parameters:

• root (string) – Root directory of dataset where airfoil_self_noise.dat exists

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

class torchhd.datasets.Annealing(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Annealing dataset.

Instances	Attributes	Task	Area
798	38	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Arrhythmia(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Arrhythmia dataset.

Instances	Attributes	Task	Area
452	279	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.AudiologyStd(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Audiology (Standardized) dataset.

Instances	Attributes	Task	Area
226	69	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.BalanceScale(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Balance Scale dataset.

Instances	Attributes	Task	Area
625	4	Classification	Social

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Balloons(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Balloons dataset.

Instances	Attributes	Task	Area
16	4	Classification	Social

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Bank(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Bank Marketing dataset.

Instances	Attributes	Task	Area
45211	17	Classification	Business

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.BeijingAirQuality(root: str, transform: Callable | None = None, download: bool = False)

Beijing Multi-Site Air-Quality dataset.

Instances	Attributes	Task	Area
420768	18	Regression	Physical

Warning

The data contains NaN values that need to be taken into account.

Parameters:

• root (string) – Root directory of dataset where directory beijing-air-quality exists or will be saved to if download is set to True.

• transform (callable, optional) – A function/transform that takes in a feature Tensor and returns a transformed version.

• download (bool, optional) – If true, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Blood(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Blood Transfusion Service Center dataset.

Instances	Attributes	Task	Area
748	5	Classification	Business

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.BreastCancer(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Breast Cancer dataset.

Instances	Attributes	Task	Area
286	9	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.BreastCancerWisc(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Breast Cancer Wisconsin (Original) dataset.

Instances	Attributes	Task	Area
699	10	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.BreastCancerWiscDiag(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Breast Cancer Wisconsin (Diagnostic) dataset.

Instances	Attributes	Task	Area
569	32	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.BreastCancerWiscProg(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Breast Cancer Wisconsin (Prognostic) dataset.

Instances	Attributes	Task	Area
198	34	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.BreastTissue(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Breast Tissue dataset.

Instances	Attributes	Task	Area
106	10	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Car(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Car Evaluation dataset.

Instances	Attributes	Task	Area
1728	6	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Cardiotocography10Clases(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Cardiotocography dataset.

Instances	Attributes	Task	Area
2126	23	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Cardiotocography3Clases(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Cardiotocography dataset.

Instances	Attributes	Task	Area
2126	23	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.ChessKrvk(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Chess (King-Rook vs. King) dataset.

Instances	Attributes	Task	Area
28056	6	Classification	Game

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.ChessKrvkp(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Chess (King-Rook vs. King-Pawn) dataset.

Instances	Attributes	Task	Area
3196	36	Classification	Game

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.CongressionalVoting(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Congressional Voting Records dataset.

Instances	Attributes	Task	Area
435	16	Classification	Social

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.ConnBenchSonarMinesRocks(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Connectionist Bench (Sonar, Mines vs. Rocks) dataset.

Instances	Attributes	Task	Area
208	60	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.ConnBenchVowelDeterding(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Connectionist Bench (Vowel Recognition - Deterding Data) dataset.

Instances	Attributes	Task	Area
528	10	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Connect4(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Connect-4 dataset.

Instances	Attributes	Task	Area
67557	42	Classification	Game

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Contrac(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Contraceptive Method Choice dataset.

Instances	Attributes	Task	Area
1473	9	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.CreditApproval(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Credit Approval dataset.

Instances	Attributes	Task	Area
690	15	Classification	Financial

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.CyclePowerPlant(root: str, download: bool = False, transform: Callable | None = None, target_transform: Callable | None = None)

Combined cycle power planet dataset.

Features consist of hourly average ambient variables Temperature (T), Ambient Pressure (AP), Relative Humidity (RH) and Exhaust Vacuum (V) to predict the net hourly electrical energy output (EP) of the plant.

Instances	Attributes	Task	Area
9568	4	Regression	Computer

Parameters:

• root (string) – Root directory of dataset where downloaded dataset exists

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

class torchhd.datasets.CylinderBands(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Cylinder Bands dataset.

Instances	Attributes	Task	Area
798	38	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Dermatology(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Dermatology dataset.

Instances	Attributes	Task	Area
366	33	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.EMGHandGestures(root: str, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False, subjects: list = [0, 1, 2, 3, 4], window: int = 256)

EMG-based hand gestures dataset.

Dataset from the paper “Hyperdimensional Biosignal Processing: A Case Study for EMG-based Hand Gesture Recognition”.

Parameters:

• root (string) – Root directory of dataset where files are stored.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• subjects (list[int], optional) – The subject numbers from 0 til 4 to include. Defaults to [0, 1, 2, 3, 4].

• window (int, optional) – The number of measurements to include in each sample. Defaults to 256.

class torchhd.datasets.Echocardiogram(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Echocardiogram dataset.

Instances	Attributes	Task	Area
132	12	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Ecoli(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Ecoli dataset.

Instances	Attributes	Task	Area
336	8	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.EnergyY1(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Energy efficiency dataset.

Instances	Attributes	Task	Area
768	8	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.EnergyY2(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Energy efficiency dataset.

Instances	Attributes	Task	Area
768	8	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.EuropeanLanguages(root: str, train: bool = True, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

European Languages dataset.

As used in the paper “A Robust and Energy-Efficient Classifier Using Brain-Inspired Hyperdimensional Computing”. The dataset contains sentences in 21 European languages, the training data was taken from Wortschatz Corpora and the testing data from Europarl Parallel Corpus.

Parameters:

• root (string) – Root directory of dataset where the training and testing samples are located.

• train (bool, optional) – If True, creates dataset from Wortschatz Corpora, otherwise from Europarl Parallel Corpus.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

• transform (callable, optional) – A function/transform that takes in an torch.LongTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

class torchhd.datasets.Fertility(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Fertility dataset.

Instances	Attributes	Task	Area
100	10	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Flags(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Flags dataset.

Instances	Attributes	Task	Area
194	30	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Glass(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Glass Identification dataset.

Instances	Attributes	Task	Area
214	10	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.HabermanSurvival(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Haberman’s Survival dataset.

Instances	Attributes	Task	Area
306	3	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.HayesRoth(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Hayes-Roth dataset.

Instances	Attributes	Task	Area
160	5	Classification	Social

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.HeartCleveland(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Heart Disease dataset.

Instances	Attributes	Task	Area
303	75	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.HeartHungarian(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Heart Disease dataset.

Instances	Attributes	Task	Area
303	75	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.HeartSwitzerland(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Heart Disease dataset.

Instances	Attributes	Task	Area
303	75	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.HeartVa(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Heart Disease dataset.

Instances	Attributes	Task	Area
303	75	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Hepatitis(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Hepatitis dataset.

Instances	Attributes	Task	Area
155	19	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.HillValley(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Hill-Valley dataset.

Instances	Attributes	Task	Area
606	101	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.HorseColic(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Horse Colic dataset.

Instances	Attributes	Task	Area
368	27	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.ISOLET(root: str, train: bool = True, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

ISOLET dataset.

Instances	Attributes	Task	Area
7797	617	Classification	Computer

Parameters:

• root (string) – Root directory of dataset where isolet1+2+3+4.data and isolet5.data exist.

• train (bool, optional) – If True, creates dataset from isolet1+2+3+4.data, otherwise from isolet5.data.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

class torchhd.datasets.IlpdIndianLiver(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

ILPD (Indian Liver Patient Dataset) dataset.

Instances	Attributes	Task	Area
583	10	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.ImageSegmentation(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Image Segmentation dataset.

Instances	Attributes	Task	Area
583	10	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Ionosphere(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Ionosphere dataset.

Instances	Attributes	Task	Area
351	34	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Iris(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Iris dataset.

Instances	Attributes	Task	Area
150	4	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.LedDisplay(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

LED Display Domain dataset.

Instances	Attributes	Task	Area
N/A	7	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Lenses(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Lenses dataset.

Instances	Attributes	Task	Area
24	4	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Letter(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Letter Recognition dataset.

Instances	Attributes	Task	Area
20000	16	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Libras(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Libras Movement dataset.

Instances	Attributes	Task	Area
360	91	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.LowResSpect(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Low Resolution Spectrometer dataset.

Instances	Attributes	Task	Area
531	102	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.LungCancer(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Lung Cancer dataset.

Instances	Attributes	Task	Area
32	56	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Lymphography(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Lymphography dataset.

Instances	Attributes	Task	Area
148	18	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Magic(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

MAGIC Gamma Telescope dataset.

Instances	Attributes	Task	Area
19020	11	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Mammographic(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Mammographic Mass dataset.

Instances	Attributes	Task	Area
961	6	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Miniboone(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

MiniBooNE particle identification dataset.

Instances	Attributes	Task	Area
130065	50	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.MolecBiolPromoter(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Molecular Biology (Promoter Gene Sequences) dataset.

Instances	Attributes	Task	Area
106	58	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.MolecBiolSplice(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Molecular Biology (Splice-junction Gene Sequences) dataset.

Instances	Attributes	Task	Area
3190	61	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Monks1(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

MONK’s Problems dataset.

Instances	Attributes	Task	Area
432	7	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Monks2(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

MONK’s Problems dataset.

Instances	Attributes	Task	Area
432	7	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Monks3(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

MONK’s Problems dataset.

Instances	Attributes	Task	Area
432	7	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Mushroom(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Mushroom dataset.

Instances	Attributes	Task	Area
8124	22	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Musk1(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Musk (Version 1) dataset.

Instances	Attributes	Task	Area
476	168	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Musk2(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Musk (Version 2) dataset.

Instances	Attributes	Task	Area
6598	168	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Nursery(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Nursery dataset.

Instances	Attributes	Task	Area
12960	8	Classification	Social

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.OocytesMerlucciusNucleus4d(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Description of the dataset is not available.

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.OocytesMerlucciusStates2f(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Description of the dataset is not available.

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.OocytesTrisopterusNucleus2f(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Description of the dataset is not available.

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.OocytesTrisopterusStates5b(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Description of the dataset is not available.

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Optical(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Optical Recognition of Handwritten Digits dataset.

Instances	Attributes	Task	Area
5620	64	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Ozone(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Ozone Level Detection dataset.

Instances	Attributes	Task	Area
2536	73	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PAMAP(root: str, subjects: list = [0, 1, 2, 3, 4, 5, 6, 7, 8], optional: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

PAMAP dataset.

Instances	Attributes	Task	Area
3850505	52	Classification	Computer

Parameters:

• root (string) – Root directory of dataset.

• subjects (list) – List of subjects to be loaded in dataset

• optional (bool) – If true optional data of some subjectes will be loaded.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

class torchhd.datasets.PageBlocks(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Page Blocks dataset.

Instances	Attributes	Task	Area
5473	10	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Parkinsons(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Parkinsons dataset.

Instances	Attributes	Task	Area
197	23	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Pendigits(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Pen-Based Recognition of Handwritten Digits dataset.

Instances	Attributes	Task	Area
10992	16	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Pima(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Pima Indians Diabetes dataset.

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PittsburgBridgesMaterial(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Pittsburgh Bridges dataset.

Instances	Attributes	Task	Area
108	13	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PittsburgBridgesRelL(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Pittsburgh Bridges dataset.

Instances	Attributes	Task	Area
5620	64	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PittsburgBridgesSpan(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Pittsburgh Bridges dataset.

Instances	Attributes	Task	Area
108	13	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PittsburgBridgesTOrD(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Pittsburgh Bridges dataset.

Instances	Attributes	Task	Area
108	13	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PittsburgBridgesType(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Pittsburgh Bridges dataset.

Instances	Attributes	Task	Area
108	13	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Planning(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Planning Relax dataset.

Instances	Attributes	Task	Area
182	13	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PlantMargin(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

One-hundred Plant Species Leaves dataset.

Instances	Attributes	Task	Area
1600	64	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PlantShape(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

One-hundred Plant Species Leaves dataset.

Instances	Attributes	Task	Area
1600	64	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PlantTexture(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

One-hundred Plant Species Leaves dataset.

Instances	Attributes	Task	Area
1600	64	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PostOperative(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Post-Operative Patient dataset.

Instances	Attributes	Task	Area
90	8	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.PrimaryTumor(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Primary Tumor dataset.

Instances	Attributes	Task	Area
399	17	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Ringnorm(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Ringnorm dataset.

Instances	Attributes	Task	Area
7400	21	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Seeds(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Seeds dataset.

Instances	Attributes	Task	Area
210	7	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Semeion(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Semeion Handwritten Digit dataset.

Instances	Attributes	Task	Area
1593	256	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Soybean(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Soybean (Large) dataset.

Instances	Attributes	Task	Area
307	35	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Spambase(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Spambase dataset.

Instances	Attributes	Task	Area
4601	57	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Spect(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

SPECT Heart Data dataset.

Instances	Attributes	Task	Area
267	22	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Spectf(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

SPECTF Heart Data dataset.

Instances	Attributes	Task	Area
267	44	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.StatlogAustralianCredit(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Statlog (Australian Credit Approval) dataset.

Instances	Attributes	Task	Area
690	14	Classification	Financial

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.StatlogGermanCredit(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Statlog (German Credit Data) dataset.

Instances	Attributes	Task	Area
1000	20	Classification	Financial

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.StatlogHeart(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Statlog (Heart) dataset.

Instances	Attributes	Task	Area
270	13	Classification	Financial

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.StatlogImage(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Statlog (Image Segmentation) dataset.

Instances	Attributes	Task	Area
2310	19	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.StatlogLandsat(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Statlog (Landsat Satellite) dataset.

Instances	Attributes	Task	Area
6435	36	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.StatlogShuttle(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Statlog (Shuttle) dataset.

Instances	Attributes	Task	Area
58000	9	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.StatlogVehicle(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Statlog (Vehicle Silhouettes) dataset.

Instances	Attributes	Task	Area
946	18	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.SteelPlates(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Steel Plates Faults dataset.

Instances	Attributes	Task	Area
1941	27	Classification	Physiscal

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.SyntheticControl(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Synthetic Control Chart Time Series dataset.

Instances	Attributes	Task	Area
600	N/A	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Teaching(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Teaching Assistant Evaluation dataset.

Instances	Attributes	Task	Area
151	5	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Thyroid(root: str, train: bool = True, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Thyroid Disease dataset.

Instances	Attributes	Task	Area
7200	21	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by hyper_search variable. Otherwise returns a subset of train dataset if hyperparameter search is performed (hyper_search = True) if not (hyper_search = False) returns test set.

• hyper_search (bool, optional) – If True, creates dataset using indices in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.TicTacToe(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Tic-Tac-Toe Endgame dataset.

Instances	Attributes	Task	Area
958	9	Classification	Game

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Titanic(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Titanic dataset.

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Trains(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Trains dataset.

Instances	Attributes	Task	Area
10	32	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Twonorm(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Leo Breiman’s twonorm example - Classification of 2 overlapping normal distributions.

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.UCIClassificationBenchmark(root: str, download: bool)

Class for iterating over all datasets used in Do we Need Hundreds of Classifiers to Solve Real World Classification Problems? from the UCI Machine Learning Repository.

Parameters:

• root (string) – Root directory containing the files of the dataset.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class DatasetEntry(name, train, test)

name: str

Alias for field number 0

test: Dataset

Alias for field number 2

train: Dataset

Alias for field number 1

datasets() → Generator[DatasetEntry, None, None]

Returns an iterator over all datasets in the benchmark.

report(dataset: DatasetEntry, metric: float) → None

Report the metric, e.g., accuracy, of the current dataset.

score() → Dict[str, List[float]]

Get the score on each dataset, averaged over cross-fold validation.

class torchhd.datasets.UCIHAR(root: str, train: bool = True, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

UCI Human Activity Recognition dataset. As found in the paper “Human Activity Recognition Using Smartphones”.

Instances	Attributes	Task	Area
10299	561	Classification	N/A

Parameters:

• root (string) – Root directory of dataset where the training and testing samples are located.

• train (bool, optional) – If True, creates dataset from UCIHAR-training data, otherwise from UCIHAR-testing data

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

• transform (callable, optional) – A function/transform that takes in an torch.LongTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

class torchhd.datasets.VertebralColumn2Clases(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Vertebral Column dataset.

Instances	Attributes	Task	Area
310	6	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.VertebralColumn3Clases(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Vertebral Column dataset.

Instances	Attributes	Task	Area
310	6	Classification	N/A

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.WallFollowing(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Wall-Following Robot Navigation dataset.

Instances	Attributes	Task	Area
5456	24	Classification	Computer

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Waveform(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Waveform Database Generator (Version 1) dataset.

Instances	Attributes	Task	Area
5000	21	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.WaveformNoise(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Waveform Database Generator (Version 2) dataset.

Instances	Attributes	Task	Area
5000	40	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Wine(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Wine dataset.

Instances	Attributes	Task	Area
178	13	Classification	Physical

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.WineQualityRed(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Wine Quality dataset.

Instances	Attributes	Task	Area
4898	12	Classification	Business

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.WineQualityWhite(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Wine Quality dataset.

Instances	Attributes	Task	Area
4898	13	Classification	Business

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Yeast(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Yeast dataset.

Instances	Attributes	Task	Area
1484	8	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

class torchhd.datasets.Zoo(root: str, train: bool = True, fold: int = -1, hyper_search: bool = False, transform: Callable | None = None, target_transform: Callable | None = None, download: bool = False)

Zoo dataset.

Instances	Attributes	Task	Area
101	17	Classification	Life

Parameters:

• root (string) – Root directory containing the files of the dataset.

• train (bool, optional) – If True, returns training (sub)set from the file storing training data as further determined by fold and hyper_search variables. Otherwise returns a subset of train dataset if hypersearch is performed (hyper_search = True) if not (hyper_search = False) returns a subset of training dataset as specified in conxuntos_kfold.dat if fold number is correct. Otherwise issues an error.

• fold (int, optional) – Specifies which fold number to use. The default value of -1 returns all the training data from the corresponding file. Values between 0 and 3 specify, which fold in conxuntos_kfold.dat to use. Relevant only if hyper_search is set to False and 0 <= fold <= 3. Indices in even rows (zero indexing) of conxuntos_kfold.dat correspond to train subsets while indices in odd rows correspond to test subsets.

• hyper_search (bool, optional) – If True, creates dataset using indeces in conxuntos.dat. This split is used for hyperparameter search. The first row corresponds to train indices (used if train = True) while the second row corresponds to test indices (used if train = False).

• transform (callable, optional) – A function/transform that takes in an torch.FloatTensor and returns a transformed version.

• target_transform (callable, optional) – A function/transform that takes in the target and transforms it.

• download (bool, optional) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.

Base classes

CollectionDataset(root[, transform, ...])	Generic class for loading datasets used in Do we Need Hundreds of Classifiers to Solve Real World Classification Problems?.
DatasetFourFold(root[, train, fold, ...])	Generic class for loading datasets without separate test data that were used in Do we Need Hundreds of Classifiers to Solve Real World Classification Problems?.
DatasetTrainTest(root[, train, ...])	Generic class for loading datasets with separate files for train and test data that were used in Do we Need Hundreds of Classifiers to Solve Real World Classification Problems?.