Maximum Entropy Baseline for Integrated Gradients
This work tackles a critical issue in AI interpretability for researchers and practitioners by providing a more consistent baseline for IG, though it is incremental as it builds on existing IG methodology.
The study addresses the ambiguity in baseline selection for Integrated Gradients (IG), a popular explainability method, by proposing a Maximum Entropy Baseline that aligns with the 'uninformative' property, and introduces an improved evaluation approach to maintain information conservativeness, with experimental assessments showing enhanced reliability in explanations.
Integrated Gradients (IG), one of the most popular explainability methods available, still remains ambiguous in the selection of baseline, which may seriously impair the credibility of the explanations. This study proposes a new uniform baseline, i.e., the Maximum Entropy Baseline, which is consistent with the "uninformative" property of baselines defined in IG. In addition, we propose an improved ablating evaluation approach incorporating the new baseline, where the information conservativeness is maintained. We explain the linear transformation invariance of IG baselines from an information perspective. Finally, we assess the reliability of the explanations generated by different explainability methods and different IG baselines through extensive evaluation experiments.