FACADE: A Framework for Adversarial Circuit Anomaly Detection and Evaluation
This work addresses adversarial attacks in deep neural networks, which is a critical security issue for AI systems, but it appears incremental as it builds on existing anomaly detection methods without claiming major breakthroughs.
The paper tackles the problem of understanding and mitigating adversarial attacks in deep neural networks by introducing FACADE, a probabilistic and geometric framework for unsupervised mechanistic anomaly detection, which aims to generate distributions over circuits to uncover adversarial attacks and improve model robustness.
We present FACADE, a novel probabilistic and geometric framework designed for unsupervised mechanistic anomaly detection in deep neural networks. Its primary goal is advancing the understanding and mitigation of adversarial attacks. FACADE aims to generate probabilistic distributions over circuits, which provide critical insights to their contribution to changes in the manifold properties of pseudo-classes, or high-dimensional modes in activation space, yielding a powerful tool for uncovering and combating adversarial attacks. Our approach seeks to improve model robustness, enhance scalable model oversight, and demonstrates promising applications in real-world deployment settings.