arnaudvl

Great, the examples you linked look like a very user friendly way indeed. Looking forward to the release!

1. Stateful here relates to whatever state (attributes) changes between prediction calls, not what state is set in the init of the detector, which would make all detectors stateful of...

Hi @KevinRyu , we just added a new backend to the MMDDrift detector ([KeOps](https://www.kernel-operations.io/keops/index.html)) which hopefully addresses your issue. We haven't released a new Alibi Detect version with it yet...

Hi @KevinRyu , just a quick note on the OOM issues. Basic PyTorch implementation of the MMD detector on your RTX3080Ti will likely get OOM from around 20k instances (so...

As far as I know, the implementations in PyOD are offline (please correct me if I'm wrong!), including for methods which could be online such as [Loda](https://link.springer.com/article/10.1007/s10994-015-5521-0). Anomaly, adversarial and...

Hi @Bennievdbuurt . While it could be fine if you only have a very limited amount of outliers during the training stage, it is not recommended because the VAE model...

Realistically, it will be a very tough problem without knowing more about what your outliers would look like and how many you could expect. Assume the following setup: you apply...

Hi @travishsu . Good point. We were thinking of moving to overriding the `train_step` (#114) method in TensorFlow so you can directly call `.fit` with all the bells and whistles....

Hi @gunja8 . You are right that the results might be slightly different as the random seeds are not fixed everywhere in the spot-the-diff detector (e.g. [here](https://github.com/SeldonIO/alibi-detect/blob/f2cca5b38902777b7cd35448929010213dc4375d/alibi_detect/cd/pytorch/spot_the_diff.py#L129)). We will have...

Could you actually verify that it isn't solved by simply setting `np.random.seed()` in your runtime?