loss_funcs module

class losses.loss_funcs.LatentCovLoss(*args: Any, **kwargs: Any)

Computes Covariance Loss on input batch.

forward(latent, _)

Parameters:: latent – is the latent space representation of the current batch
Returns:: covariance loss

class losses.loss_funcs.MAEDistLoss(*args: Any, **kwargs: Any)

Mean Absolute Error between original and latent distances

forward(distances, _)

Parameters:: distances – batch of original and latent distances between twins
Returns:: mean of absolute errors

class losses.loss_funcs.MALEDistLoss(*args: Any, **kwargs: Any)

Mean Absolute Error between logarithm of original and latent distances

__init__(factor=1.0)

forward(distances, _)

Parameters:: distances – batch of original and latent distances between twins
Returns:: mean of absolute logarithmic errors

class losses.loss_funcs.MAPEDistLoss(*args: Any, **kwargs: Any)

Mean Absolute Percentage Error between original and latent distances

forward(distances, _)

Parameters:: distances – batch of original and latent distances between twins
Returns:: mean of absolute percentage errors

class losses.loss_funcs.MSEDistLoss(*args: Any, **kwargs: Any)

Mean Squared Error between original and latent distances

forward(distances, _)

Parameters:: distances – batch of original and latent distances between twins
Returns:: mean of squared errors

class losses.loss_funcs.MSLEDistLoss(*args: Any, **kwargs: Any)

Mean Squared Error between logarithm of original and latent distances

__init__(factor=1.0)

forward(distances, _)

Parameters:: distances – batch of original and latent distances between twins
Returns:: mean of squared logarithmic errors

class losses.loss_funcs.MSPEDistLoss(*args: Any, **kwargs: Any)

Mean Squared Percentage Error between original and latent distances

forward(distances, _)

Parameters:: distances – batch of original and latent distances between twins
Returns:: mean of squared percentage errors

class losses.loss_funcs.CorrDistLoss(*args: Any, **kwargs: Any)

Correlation loss between original and latent distances

forward(distances, _)

Parameters:: distances – batch of original and latent distances between twins
Returns:: 1 - correlation coefficient

class losses.loss_funcs.CorrLogDistLoss(*args: Any, **kwargs: Any)

Correlation loss between logarithm of original and latent distances

__init__(factor=1.0)

forward(distances, _)

Parameters:: distances – batch of original and latent distances between twins
Returns:: 1 - correlation coefficient (of logarithmic distances)