alibi_detect.models.tensorflow.gmm

Functions

gmm_energy

gmm_energy(z: tensorflow.python.framework.tensor.Tensor, phi: tensorflow.python.framework.tensor.Tensor, mu: tensorflow.python.framework.tensor.Tensor, cov: tensorflow.python.framework.tensor.Tensor, L: tensorflow.python.framework.tensor.Tensor, log_det_cov: tensorflow.python.framework.tensor.Tensor, return_mean: bool = True) -> Tuple[tensorflow.python.framework.tensor.Tensor, tensorflow.python.framework.tensor.Tensor]

Compute sample energy from Gaussian Mixture Model.

Name	Type	Default	Description
z	tensorflow.python.framework.tensor.Tensor		Observations.
phi	tensorflow.python.framework.tensor.Tensor		Mixture component distribution weights.
mu	tensorflow.python.framework.tensor.Tensor		Mixture means.
cov	tensorflow.python.framework.tensor.Tensor		Mixture covariance.
L	tensorflow.python.framework.tensor.Tensor		Cholesky decomposition of cov.
log_det_cov	tensorflow.python.framework.tensor.Tensor		Log of the determinant of cov.
return_mean	bool	True	Take mean across all sample energies in a batch.

Returns

• Type: Tuple[tensorflow.python.framework.tensor.Tensor, tensorflow.python.framework.tensor.Tensor]

gmm_params

gmm_params(z: tensorflow.python.framework.tensor.Tensor, gamma: tensorflow.python.framework.tensor.Tensor) -> Tuple[tensorflow.python.framework.tensor.Tensor, tensorflow.python.framework.tensor.Tensor, tensorflow.python.framework.tensor.Tensor, tensorflow.python.framework.tensor.Tensor, tensorflow.python.framework.tensor.Tensor]

Compute parameters of Gaussian Mixture Model.

Name	Type	Default	Description
z	tensorflow.python.framework.tensor.Tensor		Observations.
gamma	tensorflow.python.framework.tensor.Tensor		Mixture probabilities to derive mixture distribution weights from.

Returns

• Type: Tuple[tensorflow.python.framework.tensor.Tensor, tensorflow.python.framework.tensor.Tensor, tensorflow.python.framework.tensor.Tensor, tensorflow.python.framework.tensor.Tensor, tensorflow.python.framework.tensor.Tensor]