Free Argumentative Exchanges for Explaining Image Classifiers

TL;DR

This paper introduces Free Argumentative eXchanges (FAXs), a novel argumentation-based framework where two agents debate to explain image classifier outputs, improving faithfulness and interpretability over traditional methods.

Contribution

The paper proposes FAXs, a new multi-agent argumentative framework for explaining image classifiers through debates, enhancing interpretability and faithfulness compared to existing explanation techniques.

Findings

FAXs outperform traditional explanation methods in faithfulness.

Empirical results show FAXs achieve high consensus and persuasion rates.

Implementations are publicly available at the provided GitHub link.

Abstract

Deep learning models are powerful image classifiers but their opacity hinders their trustworthiness. Explanation methods for capturing the reasoning process within these classifiers faithfully and in a clear manner are scarce, due to their sheer complexity and size. We provide a solution for this problem by defining a novel method for explaining the outputs of image classifiers with debates between two agents, each arguing for a particular class. We obtain these debates as concrete instances of Free Argumentative eXchanges (FAXs), a novel argumentation-based multi-agent framework allowing agents to internalise opinions by other agents differently than originally stated. We define two metrics (consensus and persuasion rate) to assess the usefulness of FAXs as argumentative explanations for image classifiers. We then conduct a number of empirical experiments showing that FAXs perform well along these metrics as well as being more faithful to the image classifiers than conventional, non-argumentative explanation methods. All our implementations can be found at https://github.com/koriavinash1/FAX.