Skip to content

MaximumMeanDiscrepancy

latent.losses.MaximumMeanDiscrepancy

Computes Maximum Mean Discrepancy (MMD) loss between conditions (y_true) in y_pred.

__init__(self, n_conditions=2, kernel_method='ms_rbf', **kwargs) special

Parameters:

Name Type Description Default
n_conditions int

Positive integer indicating number of conditions.

 2
kernel_method str

Name of kernel method to use. Can be one of the following:

  • 'ms_rbf' Multi-scale RBF kernel (Lotfollahi 2019)
  • 'rbf' Basic RBF kernel
  • 'rq' Rational Quadratic kernel
 'ms_rbf'
**kwargs

Other arguments passed to keras.losses.Loss.

 {}