Saved in:
Bibliographic Details
Main Authors: Lelièvre, Pierre, Chen, Chien-Chung
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2404.13910
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908448393789440
author Lelièvre, Pierre
Chen, Chien-Chung
author_facet Lelièvre, Pierre
Chen, Chien-Chung
contents Attribution methods are primarily designed to study input component contributions to individual model predictions. However, some research applications require a summary of attribution patterns across the entire dataset to facilitate the interpretability of the scrutinized models at a task-level rather than an instance-level. It specifically applies when the localization of important input information is supposed to be stable for a specific problem but remains unidentified among numerous components. In this paper, we present a dataset-wise attribution method called Integrated Gradient Correlation (IGC) that enables region-specific analysis by a direct summation over associated components, and further relates the sum of all attributions to a model prediction score (correlation). We demonstrate IGC on synthetic data and fMRI neural signals (NSD dataset) with the study of the representation of image features in the brain and the estimation of the visual receptive field of neural populations. The resulting IGC attributions reveal selective patterns, coherent with respective model objectives.
format Preprint
id arxiv_https___arxiv_org_abs_2404_13910
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Integrated Gradient Correlation: a Dataset-wise Attribution Method
Lelièvre, Pierre
Chen, Chien-Chung
Machine Learning
Artificial Intelligence
Attribution methods are primarily designed to study input component contributions to individual model predictions. However, some research applications require a summary of attribution patterns across the entire dataset to facilitate the interpretability of the scrutinized models at a task-level rather than an instance-level. It specifically applies when the localization of important input information is supposed to be stable for a specific problem but remains unidentified among numerous components. In this paper, we present a dataset-wise attribution method called Integrated Gradient Correlation (IGC) that enables region-specific analysis by a direct summation over associated components, and further relates the sum of all attributions to a model prediction score (correlation). We demonstrate IGC on synthetic data and fMRI neural signals (NSD dataset) with the study of the representation of image features in the brain and the estimation of the visual receptive field of neural populations. The resulting IGC attributions reveal selective patterns, coherent with respective model objectives.
title Integrated Gradient Correlation: a Dataset-wise Attribution Method
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2404.13910