对抗制
稳健性(进化)
计算机科学
渲染(计算机图形)
人工智能
语义学(计算机科学)
计算
威胁模型
计算机安全
特征(语言学)
防火墙(物理)
心理干预
机器学习
仿形(计算机编程)
数据挖掘
钥匙(锁)
因果模型
基础(证据)
桥(图论)
因果推理
不完美的
特征提取
作者
Chuancheng Shi,Shangze Li,Wenjun Lu,Wenhua Wu,Cong Wang,Zifeng Cheng,Fei Shen,Tat-Seng Chua
标识
DOI:10.48550/arxiv.2601.21900
摘要
Despite their capabilities, large foundation models (LFMs) remain susceptible to adversarial manipulation. Current defenses predominantly rely on the "locality hypothesis", suppressing isolated neurons or features. However, harmful semantics act as distributed, cross-layer circuits, rendering such localized interventions brittle and detrimental to utility. To bridge this gap, we propose \textbf{TraceRouter}, a path-level framework that traces and disconnects the causal propagation circuits of illicit semantics. TraceRouter operates in three stages: (1) it pinpoints a sensitive onset layer by analyzing attention divergence; (2) it leverages sparse autoencoders (SAEs) and differential activation analysis to disentangle and isolate malicious features; and (3) it maps these features to downstream causal pathways via feature influence scores (FIS) derived from zero-out interventions. By selectively suppressing these causal chains, TraceRouter physically severs the flow of harmful information while leaving orthogonal computation routes intact. Extensive experiments demonstrate that TraceRouter significantly outperforms state-of-the-art baselines, achieving a superior trade-off between adversarial robustness and general utility. Our code will be publicly released. WARNING: This paper contains unsafe model responses.
科研通智能强力驱动
Strongly Powered by AbleSci AI