Memory

class torchhd.structures.Memory(threshold=0.5)

Associative memory of hypervector keys and any value.

Creates a memory object.

Parameters:

threshold (float, optional) – minimal similarity between input and any hypervector in memory. Default: 0.0.

Examples:

>>> memory = structures.Memory()

__delitem__(key: VSATensor) → None

Delete the (key, value) pair from an approximate key.

Parameters:

key (VSATensor) – Hypervector key used for item lookup.

Examples:

>>> del memory[letters_hv[0]]
>>> memory[letters_hv[0]]
Exception: No elements in memory

__getitem__(key: VSATensor) → Tuple[VSATensor, Any]

Get the (key, value) pair from an approximate key.

Parameters:

key (VSATensor) – Hypervector key used for item lookup.

Examples:

>>> memory[letters_hv[0]]
(tensor([-1.,  1.,  1.,  ...,  1.,  1., -1.]), 'a')

__len__() → int

Returns the number of items in memory.

Examples:

>>> len(memory)
0

__setitem__(key: VSATensor, value: Any) → None

Set the value of an (key, value) pair from an approximate key.

Parameters:

key (VSATensor) – Hypervector key used for item lookup.

Examples:

>>> memory[letters_hv[0]] = letters[1]
>>> memory[letters_hv[0]]
(tensor([-1.,  1.,  1.,  ...,  1.,  1., -1.]), 'b')

add(key: VSATensor, value: Any) → None

Adds one (key, value) pair to memory.

Parameters:

• key (VSATensor) – Hypervector used as key for adding the key-value pair.

• value (Any) – Value to be added to the memory.

Examples:

>>> letters = list(string.ascii_lowercase)
>>> letters_hv = torchhd.random(len(letters), 10000)
>>> memory.add(letters_hv[0], letters[0])

index(key: VSATensor) → int

Returns the index of the tensor in memory from an approximate key.

Parameters:

key (VSATensor) – Hypervector key used for index lookup position.

Examples:

>>> memory.index(letters_hv[0])
>>> 0