A Human-Grounded Evaluation Benchmark for Local Explanations of Machine Learning

TL;DR

This paper introduces a human attention benchmark for evaluating model explanations in image and text domains, demonstrating its effectiveness over traditional segmentation masks and revealing user biases in subjective ratings.

Contribution

It proposes a multi-layer human attention mask benchmark for explanation evaluation and demonstrates its advantages over existing ground-truth methods.

Findings

The benchmark effectively evaluates explanations using human attention data.

Threshold-agnostic evaluation surpasses single-layer segmentation masks.

User biases influence subjective ratings of model explanations.

Abstract

Research in interpretable machine learning proposes different computational and human subject approaches to evaluate model saliency explanations. These approaches measure different qualities of explanations to achieve diverse goals in designing interpretable machine learning systems. In this paper, we propose a human attention benchmark for image and text domains using multi-layer human attention masks aggregated from multiple human annotators. We then present an evaluation study to evaluate model saliency explanations obtained using Grad-cam and LIME techniques. We demonstrate our benchmark's utility for quantitative evaluation of model explanations by comparing it with human subjective ratings and ground-truth single-layer segmentation masks evaluations. Our study results show that our threshold agnostic evaluation method with the human attention baseline is more effective than single-layer object segmentation masks to ground truth. Our experiments also reveal user biases in the subjective rating of model saliency explanations.