Gradient based attribution methods for neural networks working as classifiers use gradients of network scores. Here we discuss the practical differences between using gradients of pre-softmax scores versus post-softmax scores, and their respective advantages and disadvantages.
Cite
@article{arxiv.2306.13197,
title = {Pre or Post-Softmax Scores in Gradient-based Attribution Methods, What is Best?},
author = {Miguel Lerma and Mirtha Lucas},
journal= {arXiv preprint arXiv:2306.13197},
year = {2023}
}
Comments
8 pages, 2 figures, 2023 IEEE 13th International Conference on Pattern Recognition Systems (ICPRS)