Season：面向基于迁移的对抗攻击的频谱感知正交梯度优化方法
Season: Spectrum-Aware Orthogonal Gradient Refinement for Transfer-Based Adversarial Attacks
- Tongji University（同济大学）
- Huazhong University of Science and Technology（华中科技大学）
- Wuhan LightRead Intelligent Technology Co., Ltd.（武汉睿光智能科技有限公司）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
针对基于迁移的对抗攻击在异构架构间迁移效果差的问题，提出即插即用框架Season，通过频谱感知正交梯度优化提升攻击在CNN、ViT等目标上的迁移成功率，平均提升6.6个百分点。
AI中文摘要：
基于迁移的对抗攻击在异构架构间的迁移效果往往较差，因为卷积神经网络（CNNs）偏好局部纹理，而视觉Transformer（ViTs）依赖全局形状。我们提出Season，一种针对ImageNet上黑盒目标模型的L∞迁移攻击的频谱感知正交梯度优化框架，使用白盒替代模型。Season将每次更新分解为捕捉结构线索的低频分支和捕捉纹理的高频分支。一种低显著性引导方案将高频能量重新分配到背景区域，保留ViTs依赖的前景结构。随后的正交投影迫使纹理更新位于结构方向的正交补空间，减轻特征干扰。作为一种无需训练的即插即用包装器，Season在不修改核心的情况下增强了8种梯度稳定和输入增强攻击。在8种CNN、ViT和MLP目标上，在统一协议下，Season将迁移成功率平均提高6.6个百分点，较强基线最高提高16.0个百分点。
英文摘要：
Transfer-based adversarial attacks often transfer poorly across heterogeneous architectures because CNNs favor local textures while Vision Transformers (ViTs) rely on global shapes. We propose Season, a spectrum-aware orthogonal gradient refinement framework for L-infinity transfer attacks against black-box target models on ImageNet, using a white-box surrogate. Season decomposes each update into a low-frequency branch capturing structural cues and a high-frequency branch capturing textures. A low-saliency guidance scheme reallocates high-frequency energy to background regions, preserving foreground structures that ViTs depend on. An orthogonal projection then forces the textural update to lie in the orthogonal complement of the structural direction, mitigating feature interference. As a training-free plug-and-play wrapper, Season enhances eight gradient-stabilization and input-enhancement attacks without modifying their cores. Across eight CNN, ViT, and MLP targets, Season improves transfer success rate by 6.6 percentage points on average and up to 16.0 points over strong baselines under a unified protocol.
