resonator

torchhd.resonator(input: VSATensor, estimates: VSATensor, domains: VSATensor) → VSATensor[source]

A step of the resonator network that factorizes the input.

Given current estimates for each factor, it returns the next estimates for those factors.

Parameters:

input (VSATensor) – The hypervector to be factorized.
estimates (VSATensor) – The current estimates of the factors, typically starts as a multiset of the domain.
domains (VSATensor) – The domains of each factor containing all possible factors.

Shapes:

Input: \((*, d)\)
Estimates: \((*, n, d)\)
Domains: \((*, n, m, d)\)
Output: \((*, n, d)\)

Examples:

>>> X = torchhd.random(5, 100)
>>> Y = torchhd.random(5, 100)
>>> Z = torchhd.random(5, 100)
>>> domains = torch.stack((X, Y, Z), dim=0)
>>> domains.shape
torch.Size([3, 5, 100])
>>> x_hat = torchhd.multiset(X)
>>> y_hat = torchhd.multiset(Y)
>>> z_hat = torchhd.multiset(Z)
>>> estimates = torch.stack((x_hat, y_hat, z_hat), dim=0)
>>> estimates.shape
torch.Size([3, 100])
>>> # look at similarity of estimates with the domain
>>> torchhd.dot_similarity(estimates.unsqueeze(-2), domains).squeeze(-2)
MAPTensor([[112.,  80., 136., 106., 106.],
           [ 98., 102., 100., 110.,  74.],
           [116.,  94., 104., 112.,  82.]])
>>> # Create the combined symbol
>>> s = X[0].bind(Y[1]).bind(Z[3])
>>> # resonator step
>>> estimates = torchhd.resonator(s, estimates, domains)
>>> # look at similarity of new estimates with the domain
>>> torchhd.dot_similarity(estimates.unsqueeze(-2), domains).squeeze(-2)
MAPTensor([[100.,   8.,   8.,  -2.,  -2.],
           [-18.,  70.,  44.,  -6.,  14.],
           [  8.,   2.,   4., 100.,  -2.]])
>>> # resonator step
>>> estimates = torchhd.resonator(s, estimates, domains)
>>> # look at similarity of new estimates with the domain
>>> torchhd.dot_similarity(estimates.unsqueeze(-2), domains).squeeze(-2)
MAPTensor([[100.,   8.,   8.,  -2.,  -2.],
           [  4., 100.,  14.,   0., -16.],
           [  8.,   2.,   4., 100.,  -2.]])