Distribution-Guided Local Explanation for Black-Box Classifiers

Weijie Fu, Meng Wang, Mengnan Du, Ninghao Liu, Shijie Hao, Xia Hu

TL;DR: distribution-guided local explanation framework to provide discriminative saliency maps with easy-to-set hyper-parameters

Abstract: Existing local explanation methods provide an explanation for each decision of black-box classifiers, in the form of relevance scores of features according to their contributions. To obtain satisfying explainability, many methods introduce ad hoc constraints into the classification loss to regularize these relevance scores. However, the large information gap between the classification loss and these constraints increases the difficulty of tuning hyper-parameters. To bridge this gap, in this paper we present a simple but effective mask predictor. Specifically, we model the above constraints with a distribution controller, and integrate it with a neural network to directly guide the distribution of relevance scores. The benefit of this strategy is to facilitate the setting of involved hyper-parameters, and enable discriminative scores over supporting features. The experimental results demonstrate that our method outperforms others in terms of faithfulness and explainability. Meanwhile, it also provides effective saliency maps for explaining each decision.

Code: https://github.com/iclrlocal

Keywords: explanation, cnn, saliency map