latte.metrics.core.bundles

Module Contents

Classes

DependencyAwareMutualInformationBundle

Calculate Mutual Information Gap (MIG), Dependency-Aware Mutual Information Gap (DMIG), Dependency-Blind Mutual Information Gap (XMIG), and Dependency-Aware Latent Information Gap (DLIG) between latent vectors (z) and attributes (a).

LiadInterpolatabilityBundle

Calculate latent smoothness and monotonicity.
class DependencyAwareMutualInformationBundle(reg_dim=None, discrete=False)

Bases: latte.metrics.base.OptimizedMetricBundle

Calculate Mutual Information Gap (MIG), Dependency-Aware Mutual Information Gap (DMIG), Dependency-Blind Mutual Information Gap (XMIG), and Dependency-Aware Latent Information Gap (DLIG) between latent vectors (z) and attributes (a).

Parameters:

  • reg_dim (Optional[List], optional) – regularized dimensions, by default None Attribute a[:, i] is regularized by z[:, reg_dim[i]]. If None, a[:, i] is assumed to be regularized by z[:, i]. Note that this is the reg_dim behavior of the dependency-aware family but is different from the default reg_dim behavior of the conventional MIG.

  • discrete (bool, optional) – Whether the attributes are discrete, by default False

References

[1]

  1. Chen, X. Li, R. Grosse, and D. Duvenaud, “Isolating sources of disentanglement in variational autoencoders”, in Proceedings of the 32nd International Conference on Neural Information Processing Systems, 2018.

[2]

    1. Watcharasupat and A. Lerch, “Evaluation of Latent Space Disentanglement in the Presence of Interdependent Attributes”, in Extended Abstracts of the Late-Breaking Demo Session of the 22nd International Society for Music Information Retrieval Conference, 2021.

[3]

    1. Watcharasupat, “Controllable Music: Supervised Learning of Disentangled Representations for Music Generation”, 2021.

update_state(z, a)

Update metric states. This function append the latent vectors and attributes to the internal state lists.

Parameters:

  • z (np.ndarray, (n_samples, n_features)) – a batch of latent vectors

  • a (np.ndarray, (n_samples, n_attributes) or (n_samples,)) – a batch of attribute(s)

compute()

Compute metric values from the current state. The latent vectors and attributes in the internal states are concatenated along the sample dimension and passed to the metric function to obtain the metric values.

Returns:

A dictionary of mutual information metrics with keys [‘MIG’, ‘DMIG’, ‘XMIG’, ‘DLIG’] each mapping to a corresponding metric np.ndarray of shape (n_attributes,).

Return type:

Dict[str, np.ndarray]

class LiadInterpolatabilityBundle(reg_dim=None, liad_mode='forward', max_mode='lehmer', ptp_mode='naive', reduce_mode='attribute', liad_thresh=0.001, degenerate_val=np.nan, nanmean=True, clamp=False, p=2.0)

Bases: latte.metrics.base.OptimizedMetricBundle

Calculate latent smoothness and monotonicity.

Parameters:

  • reg_dim (Optional[List], optional) – regularized dimensions, by default None Attribute a[:, i] is regularized by z[:, reg_dim[i]]. If None, a[:, i] is assumed to be regularized by z[:, i].

  • liad_mode (str, optional) – options for calculating LIAD, by default “forward”. Only “forward” is currently supported.

  • max_mode (str, optional) – options for calculating array maximum of 2nd order LIAD, by default “lehmer”. Must be one of {“lehmer”, “naive”}. If “lehmer”, the maximum is calculated using the Lehmer mean with power p. If “naive”, the maximum is calculated using the naive array maximum. Only affects smoothness.

  • ptp_mode (str, optional) – options for calculating range of 1st order LIAD for normalization, by default “naive”. Must be either “naive” or a float value in (0.0, 1.0]. If “naive”, the range is calculated using the naive peak-to-peak range. If float, the range is taken to be the range between quantile 0.5-0.5*ptp_mode and quantile 0.5+0.5*ptp_mode. Only affects smoothness.

  • reduce_mode (str, optional) – options for reduction of the return array, by default “attribute”. Must be one of {“attribute”, “samples”, “all”, “none”}. If “all”, returns a scalar. If “attribute”, an average is taken along the sample axis and the return array is of shape (n_attributes,). If “samples”, an average is taken along the attribute axis and the return array is of shape (n_samples,). If “none”, returns a smoothness matrix of shape (n_samples, n_attributes,).

  • liad_thresh (float, optional) – threshold for ignoring noisy 1st order LIAD, by default 1e-3. Only affects monotonicity.

  • degenerate_val (float, optional) – fill value for samples with all noisy LIAD (i.e., absolute value below liad_thresh), by default np.nan. Another possible option is to set this to 0.0. Only affects monotonicity.

  • nanmean (bool, optional) – whether to ignore the NaN values in calculating the return array, by default True. Ignored if reduce_mode is “none”. If all LIAD in an axis are NaNs, the return array in that axis is filled with NaNs. Only affects monotonicity.

  • clamp (bool, optional) – Whether to clamp smoothness to [0, 1], by default False. Only affects smoothness.

  • p (float, optional) – Lehmer mean power, by default 2.0 (i.e., contraharmonic mean). Only used if max_mode == “lehmer”. Must be greater than 1.0. Only affects smoothness.

References

[1]

    1. Watcharasupat, “Controllable Music: Supervised Learning of Disentangled Representations for Music Generation”, 2021.

update_state(z, a)

Update metric states. This function append the latent vectors and attributes to the internal state lists.

Returns:

  • z (np.ndarray, (n_samples, n_interp) or (n_samples, n_features or n_attributes, n_interp)) – a batch of latent vectors

  • a (np.ndarray, (n_samples, n_interp) or (n_samples, n_attributes, n_interp)) – a batch of attribute(s)

Parameters:
compute()

Compute metric values from the current state. The latent vectors and attributes in the internal states are concatenated along the sample dimension and passed to the metric function to obtain the metric values.

Returns:

A dictionary of LIAD-based interpolatability metrics with keys [‘smoothness’, ‘monotonicity’] each mapping to a corresponding metric np.ndarray. See reduce_mode for details on the shape of the return arrays.

Return type:

Dict[str, np.ndarray]