Explainability-Guided Defense: Attribution-Aware Model Refinement Against Adversarial Data Attacks

2026-01-02

arXiv.org

AI Summary

This work proposes an interpretability-guided adversarial training framework to address the tension between robustness and interpretability in deep learning models deployed in safety-critical settings. For the first time, LIME-based feature attribution is directly integrated into the training process, where feature masking and sensitivity-aware regularization actively suppress spurious or unstable features identified by the explainer—without requiring additional data or architectural modifications. Theoretical analysis establishes a lower bound linking alignment of feature attributions to model robustness. Extensive experiments demonstrate that the proposed method significantly enhances both adversarial robustness and out-of-distribution generalization on CIFAR-10, CIFAR-10-C, and CIFAR-100 benchmarks.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness
• Natural Language Processing: Safety and Robustness

Application Category

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

Abstract

The growing reliance on deep learning models in safety-critical domains such as healthcare and autonomous navigation underscores the need for defenses that are both robust to adversarial perturbations and transparent in their decision-making. In this paper, we identify a connection between interpretability and robustness that can be directly leveraged during training. Specifically, we observe that spurious, unstable, or semantically irrelevant features identified through Local Interpretable Model-Agnostic Explanations (LIME) contribute disproportionately to adversarial vulnerability. Building on this insight, we introduce an attribution-guided refinement framework that transforms LIME from a passive diagnostic into an active training signal. Our method systematically suppresses spurious features using feature masking, sensitivity-aware regularization, and adversarial augmentation in a closed-loop refinement pipeline. This approach does not require additional datasets or model architectures and integrates seamlessly into standard adversarial training. Theoretically, we derive an attribution-aware lower bound on adversarial distortion that formalizes the link between explanation alignment and robustness. Empirical evaluations on CIFAR-10, CIFAR-10-C, and CIFAR-100 demonstrate substantial improvements in adversarial robustness and out-of-distribution generalization.

Problem

• adversarial attacks
• model robustness
• explainability
• interpretability
• deep learning

Innovation

• Explainability-Guided Defense
• Attribution-Aware Refinement
• Adversarial Robustness
• LIME
• Feature Suppression