Restricting the Flow: Information Bottlenecks for AttributionDownload PDF

Published: 20 Dec 2019, Last Modified: 28 May 2023ICLR 2020 Conference Blind SubmissionReaders: Everyone
Original Pdf: pdf
Community Implementations: [![CatalyzeX](/images/catalyzex_icon.svg) 3 code implementations](
TL;DR: We apply the informational bottleneck concept to attribution.
Abstract: Attribution methods provide insights into the decision-making of machine learning models like artificial neural networks. For a given input sample, they assign a relevance score to each individual input variable, such as the pixels of an image. In this work, we adopt the information bottleneck concept for attribution. By adding noise to intermediate feature maps, we restrict the flow of information and can quantify (in bits) how much information image regions provide. We compare our method against ten baselines using three different metrics on VGG-16 and ResNet-50, and find that our methods outperform all baselines in five out of six settings. The method’s information-theoretic foundation provides an absolute frame of reference for attribution values (bits) and a guarantee that regions scored close to zero are not necessary for the network's decision.
Keywords: Attribution, Informational Bottleneck, Interpretable Machine Learning, Explainable AI
13 Replies