Neural Networks with Feature Attribution and Contrastive Explanations
Abstract
Interpretability is becoming an expected and even essential characteristic in GDPR Europe. In the majority of existing work on natural language processing (NLP), interpretability has focused on the problem of explanatory responses to questions like “ Why p? ” (identifying the causal attributes that support the prediction of " p .)” This type of local explainability focuses on explaining a single prediction made by a model for a single input, by quantifying the contribution of each feature to the predicted output class. Most of these methods are based on post-hoc approaches. In this paper, we propose a technique to learn centroid vectors concurrently while building the black-box in order to support answers to “ Why p? ” and “ Why p and not q? ,” where “ q ” is another class that is contrastive to “ p .” Across multiple datasets, our approach achieves better results than traditional post-hoc methods.
Cite
Text
Babiker et al. "Neural Networks with Feature Attribution and Contrastive Explanations." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2022. doi:10.1007/978-3-031-26387-3_23Markdown
[Babiker et al. "Neural Networks with Feature Attribution and Contrastive Explanations." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2022.](https://mlanthology.org/ecmlpkdd/2022/babiker2022ecmlpkdd-neural/) doi:10.1007/978-3-031-26387-3_23BibTeX
@inproceedings{babiker2022ecmlpkdd-neural,
title = {{Neural Networks with Feature Attribution and Contrastive Explanations}},
author = {Babiker, Housam Khalifa Bashier and Kim, Mi-Young and Goebel, Randy},
booktitle = {European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases},
year = {2022},
pages = {372-388},
doi = {10.1007/978-3-031-26387-3_23},
url = {https://mlanthology.org/ecmlpkdd/2022/babiker2022ecmlpkdd-neural/}
}