CAT: Concept-level backdoor ATtacks for Concept Bottleneck Models

TL;DR

This paper introduces CAT and CAT+, novel concept-level backdoor attack methods for Concept Bottleneck Models, revealing security vulnerabilities and providing evaluation tools to measure attack effectiveness and stealthiness.

Contribution

It presents the first concept-level backdoor attack framework for CBMs, including an enhanced version with optimized trigger selection, and offers a comprehensive evaluation methodology.

Findings

High attack success rates with minimal impact on clean data

Effective and stealthy concept trigger selection via correlation function

Demonstrated security risks in CBMs and the need for robust defenses

Abstract

Despite the transformative impact of deep learning across multiple domains, the inherent opacity of these models has driven the development of Explainable Artificial Intelligence (XAI). Among these efforts, Concept Bottleneck Models (CBMs) have emerged as a key approach to improve interpretability by leveraging high-level semantic information. However, CBMs, like other machine learning models, are susceptible to security threats, particularly backdoor attacks, which can covertly manipulate model behaviors. Understanding that the community has not yet studied the concept level backdoor attack of CBM, because of "Better the devil you know than the devil you don't know.", we introduce CAT (Concept-level Backdoor ATtacks), a methodology that leverages the conceptual representations within CBMs to embed triggers during training, enabling controlled manipulation of model predictions at inference time. An enhanced attack pattern, CAT+, incorporates a correlation function to systematically select the most effective and stealthy concept triggers, thereby optimizing the attack's impact. Our comprehensive evaluation framework assesses both the attack success rate and stealthiness, demonstrating that CAT and CAT+ maintain high performance on clean data while achieving significant targeted effects on backdoored datasets. This work underscores the potential security risks associated with CBMs and provides a robust testing methodology for future security assessments.