Spurious Correlations in Concept Drift: Can Explanatory Interaction Help?

TL;DR

This paper introduces ebc-exstream, a novel concept drift detector that uses model explanations and human feedback to identify and mitigate spurious correlations, improving detection accuracy in long-term machine learning models.

Contribution

The paper proposes ebc-exstream, a new detection method leveraging explanations and feedback to address spurious correlations in concept drift detection.

Findings

Preliminary experiments show ebc-exstream reduces spurious correlation impact.

The method uses entropy heuristics to minimize human annotation.

Initial results indicate promise in confounded data scenarios.

Abstract

Long-running machine learning models face the issue of concept drift (CD), whereby the data distribution changes over time, compromising prediction performance. Updating the model requires detecting drift by monitoring the data and/or the model for unexpected changes. We show that, however, spurious correlations (SCs) can spoil the statistics tracked by detection algorithms. Motivated by this, we introduce ebc-exstream, a novel detector that leverages model explanations to identify potential SCs and human feedback to correct for them. It leverages an entropy-based heuristic to reduce the amount of necessary feedback, cutting annotation costs. Our preliminary experiments on artificially confounded data highlight the promise of ebc-exstream for reducing the impact of SCs on detection.