H2ST: Hierarchical Two-sample Tests for Continual Out-of-Distribution Detection

Description

This repository is for the implementation of our paper, H2ST: Hierarchical Two-sample Tests for Continual Out-of-Distribution Detection

Run the code

1. Set up the environment:

pip install -r requirements.txt

2. Prepare the datasets (take MNIST as an example):

python data/mnist.py

3. Run the training and detection:

python main.py 

Credits

These open source projects played a pivotal role in our research:

• https://github.com/facebookresearch/GradientEpisodicMemory
• https://github.com/sooyongj/sequential_covariate_shift_detection
• and many more...

More details to be updated