Subspace Defense: Discarding Adversarial Perturbations by Learning a Subspace for Clean Signals
원문 초록
Deep neural networks (DNNs) are notoriously vulnerable to adversarial attacks that place carefully crafted perturbations on normal examples to fool DNNs. To better understand such attacks, a characterization of the features carried by adversarial examples is needed. In this paper, we tackle this challenge by inspecting the subspaces of sample features through spectral analysis. We first empirically show that the features of either clean signals or adversarial perturbations are redundant and span in low-dimensional linear subspaces respectively with minimal overlap, and the classical low-dimensional subspace projection can suppress perturbation features out of the subspace of clean signals. This makes it possible for DNNs to learn a subspace where only features of clean signals exist while those of perturbations are discarded, which can facilitate the distinction of adversarial examples. To prevent the residual perturbations that is inevitable in subspace learning, we propose an independence criterion to disentangle clean signals from perturbations. Experimental results show that the proposed strategy enables the model to inherently suppress adversaries, which not only boosts model robustness but also motivates new directions of effective adversarial defense.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Bridging Robustness and Generalization Against Word Substitution Attacks in NLP via the Growth Bound Matrix Approach
Towards Effective and Efficient Adversarial Defense with Diffusion Models for Robust Visual Tracking
Accidental Misalignment: Fine-Tuning Language Models Induces Unexpected Vulnerability
A Few Large Shifts: Layer-Inconsistency Based Minimal Overhead Adversarial Example Detection
GenoArmory: A Unified Evaluation Framework for Adversarial Attacks on Genomic Foundation Models
Lightweight Defense Against Adversarial Attacks in Time Series Classification
토론
이 논문과 연결된 포럼 글이 아직 없습니다.