A Dual-Perspective Approach to Evaluating Feature Attribution Methods

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Yawei, Zhang, Yang, Kawaguchi, Kenji, Khakzar, Ashkan, Bischl, Bernd, Rezaei, Mina
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929602673246208
author Li, Yawei
Zhang, Yang
Kawaguchi, Kenji
Khakzar, Ashkan
Bischl, Bernd
Rezaei, Mina
author_facet Li, Yawei
Zhang, Yang
Kawaguchi, Kenji
Khakzar, Ashkan
Bischl, Bernd
Rezaei, Mina
contents Feature attribution methods attempt to explain neural network predictions by identifying relevant features. However, establishing a cohesive framework for assessing feature attribution remains a challenge. There are several views through which we can evaluate attributions. One principal lens is to observe the effect of perturbing attributed features on the model's behavior (i.e., faithfulness). While providing useful insights, existing faithfulness evaluations suffer from shortcomings that we reveal in this paper. In this work, we propose two new perspectives within the faithfulness paradigm that reveal intuitive properties: soundness and completeness. Soundness assesses the degree to which attributed features are truly predictive features, while completeness examines how well the resulting attribution reveals all the predictive features. The two perspectives are based on a firm mathematical foundation and provide quantitative metrics that are computable through efficient algorithms. We apply these metrics to mainstream attribution methods, offering a novel lens through which to analyze and compare feature attribution methods.
format Preprint
id arxiv_https___arxiv_org_abs_2308_08949
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle A Dual-Perspective Approach to Evaluating Feature Attribution Methods
Li, Yawei
Zhang, Yang
Kawaguchi, Kenji
Khakzar, Ashkan
Bischl, Bernd
Rezaei, Mina
Machine Learning
Artificial Intelligence
Feature attribution methods attempt to explain neural network predictions by identifying relevant features. However, establishing a cohesive framework for assessing feature attribution remains a challenge. There are several views through which we can evaluate attributions. One principal lens is to observe the effect of perturbing attributed features on the model's behavior (i.e., faithfulness). While providing useful insights, existing faithfulness evaluations suffer from shortcomings that we reveal in this paper. In this work, we propose two new perspectives within the faithfulness paradigm that reveal intuitive properties: soundness and completeness. Soundness assesses the degree to which attributed features are truly predictive features, while completeness examines how well the resulting attribution reveals all the predictive features. The two perspectives are based on a firm mathematical foundation and provide quantitative metrics that are computable through efficient algorithms. We apply these metrics to mainstream attribution methods, offering a novel lens through which to analyze and compare feature attribution methods.
title A Dual-Perspective Approach to Evaluating Feature Attribution Methods
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2308.08949