Benchmarking and Survey of Explanation Methods for Black Box Models

Rapporteurs: Francesco Bodria, Francesca Naretto; Guest: Muhammad Rehman Zafar

Rationale:
The widespread adoption of black box models in artificial intelligence has increased the need for explanatory methods to reveal how these obscure models reach specific
decisions. Retrieving explanations is crucial to unveil possible biases and resolve practical or ethical issues. Nowadays, the literature is full of methods with different explanations. This study shows a visual comparison of explanations and a quantitative benchmarking of various explainers. The full survey is available here: https://arxiv.org/abs/2102.13076