arnaudvl

Add KeOps MMD detector

13

Add MMD detector using the [KeOps](https://github.com/getkeops/keops) (PyTorch) backend to further accelerate drift detection and scale up to larger datasets. This PR needs to be made compatible with the optional dependency...

Ensure that PyTorch, TensorFlow and KeOps kernels all take native tensors from the relevant backend as inputs. Currently PyTorch allows to pass NumPy arrays in the forward pass but doesn't...

Allow non-tensorflow classifiers in ClassifierDrift

2

Extend method for eg scikit-learn or xgboost classifiers.

Currently loading a saved drift detector which uses a preprocessing step via a `preprocess_fn` [couples the preprocessing step to the presence of a model](https://github.com/SeldonIO/alibi-detect/blob/master/alibi_detect/utils/saving.py#L1503), which should not be necessary. This...

Clarify when `list` input types for the drift detectors require a `preprocess_fn` as well. E.g. a `list` input cannot occur in the `KSDrift` or `ChiSquareDrift` detectors without additional `preprocess_fn` while...

Correct type declaration of numpy types

1

Check e.g. in #278 : assume an input argument of a class or function is `dtype: TYPE = np.float32`. `TYPE` should not be `np.dtype` as the numpy types such as...