Generative Example-Based Explanations: Bridging the Gap between Generative Modeling and Explainability

2024-10-28

arXiv.org

AI Summary

Existing exemplar-based image explanation methods leveraging deep generative models deliver visually compelling outputs but suffer from a significant conceptual and evaluative disconnect with the broader eXplainable AI (XAI) community. Method: We propose the first unified probabilistic framework that formally encodes classical explanation desiderata—faithfulness, locality, and succinctness—as generative modeling objectives, realized through Bayesian inference and feature-semantic alignment to yield semantically interpretable and theoretically verifiable local exemplar explanations in high-dimensional spaces. Our approach is architecture-agnostic, seamlessly integrating with mainstream generative models including VAEs and GANs. Contribution/Results: The framework systematically bridges the conceptual gap between generative explanations and conventional XAI paradigms. Empirically, it achieves substantial improvements in explanation fidelity, stability, and human interpretability across image and multimodal tasks, outperforming state-of-the-art methods on multiple quantitative metrics and user studies.

Technology Category

• Humans and AI: Explainable AI (XAI) for Human Understanding
• Computer Vision: Interpretability, Explainability, and Transparency
• Machine Learning: Transparent, Interpretable, Explainable ML

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Recently, several methods have leveraged deep generative modeling to produce example-based explanations of decision algorithms for high-dimensional input data. Despite promising results, a disconnect exists between these methods and the classical explainability literature, which focuses on lower-dimensional data with semantically meaningful features. This conceptual and communication gap leads to misunderstandings and misalignments in goals and expectations. In this paper, we bridge this gap by proposing a novel probabilistic framework for local example-based explanations. Our framework integrates the critical characteristics of classical local explanation desiderata while being amenable to high-dimensional data and their modeling through deep generative models. Our aim is to facilitate communication, foster rigor and transparency, and improve the quality of peer discussion and research progress.

Problem

• Bridging generative modeling and explainability research gap
• Formally defining probabilistic example-based explanation framework
• Aligning generative methods with explainability desiderata and characteristics

Innovation

• Probabilistic framework for example-based explanations
• Formally defining explanations via generative models
• Aligning generative methods with explainability desiderata