Back to Search
Start Over
CLEVR-XAI: A benchmark dataset for the ground truth evaluation of neural network explanations.
- Source :
-
Information Fusion . May2022, Vol. 81, p14-40. 27p. - Publication Year :
- 2022
-
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. • First objective, ground truth-based evaluation framework for XAI methods. • Systematic comparison of 10 popular XAI methods using novel evaluation metrics. • New findings regarding strengths and limitations of particular XAI methods. • Evaluation framework, i.e., code and dataset, are publicly available. [ABSTRACT FROM AUTHOR]
- Subjects :
- *DEEP learning
*COMPUTER vision
*EXPLANATION
*CONVOLUTIONAL neural networks
Subjects
Details
- Language :
- English
- ISSN :
- 15662535
- Volume :
- 81
- Database :
- Academic Search Index
- Journal :
- Information Fusion
- Publication Type :
- Academic Journal
- Accession number :
- 154736081
- Full Text :
- https://doi.org/10.1016/j.inffus.2021.11.008