FACADE: A Framework for Adversarial Circuit Anomaly Detection and Evaluation

TL;DR

FACADE introduces a probabilistic and geometric framework for unsupervised detection of adversarial circuit anomalies in deep neural networks, enhancing robustness and interpretability.

Contribution

It is the first to combine probabilistic and geometric methods for unsupervised anomaly detection in neural circuits, focusing on adversarial attack mitigation.

Findings

Provides insights into circuit contributions to manifold changes

Improves model robustness against adversarial attacks

Demonstrates effectiveness in real-world scenarios

Abstract

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.