Transferable Backdoor Attack

Transferable backdoor attacks exploit vulnerabilities in machine learning models, allowing malicious actors to manipulate model outputs using hidden triggers that transfer across different models or tasks. Current research focuses on understanding the mechanisms of these attacks across various architectures, including large language models (LLMs), pre-trained models (PTMs), and graph neural networks (GNNs), and developing methods to inject and detect these backdoors, often leveraging techniques like low-rank adaptation (LoRA) or embedding space analysis. The ability of these attacks to compromise seemingly secure models highlights a critical need for robust defenses and underscores the importance of developing more resilient and trustworthy AI systems.

Papers

May 25, 2024

Breaking the False Sense of Security in Backdoor Defense through Re-Activation Attack
Mingli Zhu, Siyuan Liang, Baoyuan Wu
Backdoor Attack Security Related Backdoor Defense False Sense Transferable Backdoor Attack

February 29, 2024

LoRA-as-an-Attack! Piercing LLM Safety Under The Share-and-Play Scenario
Hongyi Liu, Zirui Liu, Ruixiang Tang, Jiayi Yuan, Shaochen Zhong, Yu-Neng Chuang, Li Li, Rui Chen, Xia Hu
LLM Safety LoRA Fine Tuning LoRA Based Transferable Backdoor Attack Collaborative Game

January 29, 2024

Model Supply Chain Poisoning: Backdooring Pre-trained Models via Embedding Indistinguishability
Hao Wang, Shangwei Guo, Jialing He, Hangcheng Liu, Tianwei Zhang, Tao Xiang
Pre Trained Model Backdoor Attack Outcome Indistinguishability Task Agnostic Backdoor Transferable Backdoor Attack

November 29, 2023

TARGET: Template-Transferable Backdoor Attack Against Prompt-based NLP Models via GPT4
Zihao Tan, Qingliang Chen, Yongjian Huang, Chen Liang
GPT 4 Prompt Based Learning Transferable Attack Target Amazon Ally Prompt Based NLP Transferable Backdoor Attack Data Free Black Box Attack

May 28, 2023

NOTABLE: Transferable Backdoor Attacks Against Prompt-based NLP Models
Kai Mei, Zheng Li, Zhenting Wang, Yang Zhang, Shiqing Ma
Backdoor Attack Prompt Based Learning Transferable Backdoor Attack Prompt Based NLP

June 21, 2022

Transferable Graph Backdoor Attack
Shuiqiao Yang, Bao Gia Doan, Paul Montague, Olivier De Vel, Tamas Abraham, Seyit Camtepe, Damith C. Ranasinghe, Salil S. Kanhere
Graph Neural Network Adversarial Attack Adversarial Example Graph Backdoor Attack Transferable Backdoor Attack

November 22, 2021

NTD: Non-Transferability Enabled Backdoor Detection
Yinshan Li, Hua Ma, Zhi Zhang, Yansong Gao, Alsharif Abuadbba, Anmin Fu, Yifeng Zheng, Said F. Al-Sarawi, Derek Abbott
Backdoor Attack Task Transferability Backdoor Detection Adversarial Trigger Transferable Backdoor Attack