Explainability for Machine Learning Models: From Data Adaptability to User Perception

TL;DR

This thesis investigates local explanation methods for deployed machine learning models, focusing on improving explanation fidelity, evaluating explanation suitability, and understanding user perception to enhance AI transparency and trust.

Contribution

It introduces a novel evaluation approach for linear explanations and compares counterfactual methods, alongside user studies on explanation impact and representation.

Findings

Enhanced rule-based explanation method

Proposed a new linear explanation evaluation approach

User experiments show explanation type affects understanding and trust

Abstract

This thesis explores the generation of local explanations for already deployed machine learning models, aiming to identify optimal conditions for producing meaningful explanations considering both data and user requirements. The primary goal is to develop methods for generating explanations for any model while ensuring that these explanations remain faithful to the underlying model and comprehensible to the users. The thesis is divided into two parts. The first enhances a widely used rule-based explanation method. It then introduces a novel approach for evaluating the suitability of linear explanations to approximate a model. Additionally, it conducts a comparative experiment between two families of counterfactual explanation methods to analyze the advantages of one over the other. The second part focuses on user experiments to assess the impact of three explanation methods and two distinct representations. These experiments measure how users perceive their interaction with the model in terms of understanding and trust, depending on the explanations and representations. This research contributes to a better explanation generation, with potential implications for enhancing the transparency, trustworthiness, and usability of deployed AI systems.