Towards Automatic Concept-based Explanations

TL;DR

This paper introduces ACE, a new algorithm for automatically extracting human-understandable visual concepts from machine learning models, aiming to improve interpretability by summarizing feature importance at a higher conceptual level.

Contribution

The paper proposes principles for concept-based explanations and develops ACE, an algorithm that automatically identifies meaningful visual concepts across datasets.

Findings

ACE discovers human-meaningful concepts

Concepts are coherent and relevant to predictions

Method enhances interpretability of neural networks

Abstract

Interpretability has become an important topic of research as more machine learning (ML) models are deployed and widely used to make important decisions. Most of the current explanation methods provide explanations through feature importance scores, which identify features that are important for each individual input. However, how to systematically summarize and interpret such per sample feature importance scores itself is challenging. In this work, we propose principles and desiderata for \emph{concept} based explanation, which goes beyond per-sample features to identify higher-level human-understandable concepts that apply across the entire dataset. We develop a new algorithm, ACE, to automatically extract visual concepts. Our systematic experiments demonstrate that \alg discovers concepts that are human-meaningful, coherent and important for the neural network's predictions.