Ground Truth Evaluation of Neural Network Explanations with CLEVR-XAI

TL;DR

This paper introduces a ground truth evaluation framework for neural network explanations in computer vision, using the CLEVR-XAI dataset, to objectively assess and compare explanation methods.

Contribution

It presents a novel, controlled, and realistic evaluation framework for XAI methods based on the CLEVR visual question answering task, enabling objective comparison.

Findings

Different explanation methods vary significantly in quality.

Some methods outperform others in specific scenarios.

The framework reveals insights contradicting previous qualitative assessments.

Abstract

The rise of deep learning in today's applications entailed an increasing need in explaining the model's decisions beyond prediction performances in order to foster trust and accountability. Recently, the field of explainable AI (XAI) has developed methods that provide such explanations for already trained neural networks. In computer vision tasks such explanations, termed heatmaps, visualize the contributions of individual pixels to the prediction. So far XAI methods along with their heatmaps were mainly validated qualitatively via human-based assessment, or evaluated through auxiliary proxy tasks such as pixel perturbation, weak object localization or randomization tests. Due to the lack of an objective and commonly accepted quality measure for heatmaps, it was debatable which XAI method performs best and whether explanations can be trusted at all. In the present work, we tackle the problem by proposing a ground truth based evaluation framework for XAI methods based on the CLEVR visual question answering task. Our framework provides a (1) selective, (2) controlled and (3) realistic testbed for the evaluation of neural network explanations. We compare ten different explanation methods, resulting in new insights about the quality and properties of XAI methods, sometimes contradicting with conclusions from previous comparative studies. The CLEVR-XAI dataset and the benchmarking code can be found at https://github.com/ahmedmagdiosman/clevr-xai.