Interpretable Distribution Shift Detection using Optimal Transport
This addresses the need for interpretable methods to analyze distribution shifts in datasets, but it is incremental as it builds on existing optimal transport techniques.
The authors tackled the problem of detecting and characterizing distribution shifts in classification datasets by proposing a method based on optimal transport, which identifies affected classes and retrieves sample pairs for insights, though results are preliminary.
We propose a method to identify and characterize distribution shifts in classification datasets based on optimal transport. It allows the user to identify the extent to which each class is affected by the shift, and retrieves corresponding pairs of samples to provide insights on its nature. We illustrate its use on synthetic and natural shift examples. While the results we present are preliminary, we hope that this inspires future work on interpretable methods for analyzing distribution shifts.