Implicit Mixture of Interpretable Experts for Global and Local Interpretability

TL;DR

This paper introduces IMoIE, a model that combines interpretable experts with a black-box router to achieve accurate and interpretable image classification, addressing issues of cheating and enabling analysis of interpretability at different scales.

Contribution

The paper proposes a novel implicit parameterization scheme for mixtures of interpretable experts, improving interpretability and performance in image classification tasks.

Findings

IMoIE matches state-of-the-art accuracy on MNIST10.

IMoIE provides local interpretability for individual decisions.

Global interpretability is achievable with some accuracy trade-offs.

Abstract

We investigate the feasibility of using mixtures of interpretable experts (MoIE) to build interpretable image classifiers on MNIST10. MoIE uses a black-box router to assign each input to one of many inherently interpretable experts, thereby providing insight into why a particular classification decision was made. We find that a naively trained MoIE will learn to 'cheat', whereby the black-box router will solve the classification problem by itself, with each expert simply learning a constant function for one particular class. We propose to solve this problem by introducing interpretable routers and training the black-box router's decisions to match the interpretable router. In addition, we propose a novel implicit parameterization scheme that allows us to build mixtures of arbitrary numbers of experts, allowing us to study how classification performance, local and global interpretability vary as the number of experts is increased. Our new model, dubbed Implicit Mixture of Interpretable Experts (IMoIE) can match state-of-the-art classification accuracy on MNIST10 while providing local interpretability, and can provide global interpretability albeit at the cost of reduced classification accuracy.