Variational Auto-Encoder (VAE)#

class oats.models.reconstruction.vae.VAEModel(window=10, **kwargs)[source]#

Bases: PyODModel

VAE Model (Variational Auto-encover)

Using reconstruction error of the trained encoder-decoder network as anomaly scores.

Reference: https://pyod.readthedocs.io/en/latest/_modules/pyod/models/vae.html

initialization also accepts any parameters used by: https://pyod.readthedocs.io/en/latest/_modules/pyod/models/vae.html

Parameters:

window (int, optional) – _description_. Defaults to 10.

fit(train_data, **kwargs)[source]#
get_scores(test_data: ndarray[Any, dtype[Any]], normalize=False)#