torch-mfa
torch-mfa copied to clipboard
Train a Mixture of Factor Analyzers (MFA) / Mixture of Probabilistic PCA (MPPCA) - low-rank-plus-diagonal GMMs using pytorch
Results
2
torch-mfa issues
Sort by
recently updated
recently updated
newest added
Dear @eitanrich, It seems there is a missing logarithm in the function `_component_log_likelihood` in mfa.py ([line 103](https://github.com/eitanrich/torch-mfa/blob/8b0b287eaaf1e9dd9d895e8065934a21a2c09512/mfa.py#L103)). Currently, it is written as: ```python return PI.reshape(1, K) + log_prob_data_given_components.T ``` I...