DeepFaith: A Domain-Free and Model-Agnostic Unified Framework for Highly Faithful Explanations

TL;DR

DeepFaith introduces a unified, model-agnostic framework that optimizes faithfulness across multiple metrics, providing highly faithful explanations without model access, and demonstrating superior performance across diverse tasks.

Contribution

It proposes a theoretically grounded, domain-free, and model-agnostic explanation framework that unifies multiple faithfulness metrics for optimal explanation quality.

Findings

Achieves highest faithfulness across 10 metrics on 12 tasks

Operates without model access after training

Demonstrates cross-domain generalizability

Abstract

Explainable AI (XAI) builds trust in complex systems through model attribution methods that reveal the decision rationale. However, due to the absence of a unified optimal explanation, existing XAI methods lack a ground truth for objective evaluation and optimization. To address this issue, we propose Deep architecture-based Faith explainer (DeepFaith), a domain-free and model-agnostic unified explanation framework under the lens of faithfulness. By establishing a unified formulation for multiple widely used and well-validated faithfulness metrics, we derive an optimal explanation objective whose solution simultaneously achieves optimal faithfulness across these metrics, thereby providing a ground truth from a theoretical perspective. We design an explainer learning framework that leverages multiple existing explanation methods, applies deduplicating and filtering to construct high-quality supervised explanation signals, and optimizes both pattern consistency loss and local correlation to train a faithful explainer. Once trained, DeepFaith can generate highly faithful explanations through a single forward pass without accessing the model being explained. On 12 diverse explanation tasks spanning 6 models and 6 datasets, DeepFaith achieves the highest overall faithfulness across 10 metrics compared to all baseline methods, highlighting its effectiveness and cross-domain generalizability.