Composite Attack

Composite attacks, encompassing multiple simultaneous attacks targeting machine learning models, are a growing area of research focusing on understanding and mitigating their effectiveness. Current work investigates various attack types, including combinations of backdoor attacks, data poisoning, adversarial examples (e.g., ℓ∞ and spatial perturbations), and instruction manipulation in large language models (LLMs), often employing generative adversarial networks (GANs) or multi-fidelity evaluation methods for attack generation and defense. Understanding and defending against these composite attacks is crucial for ensuring the security and reliability of machine learning systems across diverse applications, from image recognition to natural language processing.

Papers

June 23, 2024

CBPF: Filtering Poisoned Data Based on Composite Backdoor Attack
Hanfeng Xia, Haibo Hong, Ruili Wang
Backdoor Attack Backdoor Poisoning Attack Poisoned Data Composite Attack Target Class

November 27, 2023

A Survey on Vulnerability of Federated Learning: A Learning Algorithm Perspective
Xianghua Xie, Chen Hu, Hanchi Ren, Jingjing Deng
Timely Survey Learning Algorithm Threat Model Latent Vulnerability Malicious Client Composite Attack Malicious Attack

October 16, 2023

Prompt Packer: Deceiving LLMs through Compositional Instruction with Hidden Attacks
Shuyu Jiang, Xingshu Chen, Rui Tang
Large Language Model Stealthy Attack Unsafe Prompt Composite Attack Instruction Backdoor Attack

October 11, 2023

Composite Backdoor Attacks Against Large Language Models
Hai Huang, Zhengyu Zhao, Michael Backes, Yun Shen, Yang Zhang
Large Language Model Natural Language Processing Backdoor Attack Composite Attack

May 12, 2023

Efficient Search of Comprehensively Robust Neural Architectures via Multi-fidelity Evaluation
Jialiang Sun, Wen Yao, Tingsong Jiang, Xiaoqian Chen
Multi Fidelity Robust Deep Semantic Adversarial Efficient Search Robust Neural Architecture Composite Attack Robust Architecture

March 22, 2023

November 20, 2022

Deep Composite Face Image Attacks: Generation, Vulnerability and Detection
Jag Mohan Singh, Raghavendra Ramachandra
Data Detection Faithful Generation Latent Vulnerability Morphing Attack Facial Data Face Morphing Attack Detection Composite Attack

October 5, 2022

A Closer Look at Robustness to L-infinity and Spatial Perturbations and their Composition
Luke Rowe, Benjamin Thérien, Krzysztof Czarnecki, Hongyang Zhang
Native Robustness Adversarial Learning Glance Annotation Compositional Ability Norm Attack Composite Attack Compositional Adversarial

July 16, 2022

CARBEN: Composite Adversarial Robustness Benchmark
Lei Hsiung, Yun-Yun Tsai, Pin-Yu Chen, Tsung-Yi Ho
Adversarial Attack Adversarial Robustness Robustness Evaluation Threat Model Composite Attack

February 9, 2022

Towards Compositional Adversarial Robustness: Generalizing Adversarial Training to Composite Semantic Perturbations
Lei Hsiung, Yun-Yun Tsai, Pin-Yu Chen, Tsung-Yi Ho
Adversarial Example Adversarial Training Model Robustness Composite Attack Semantic Perturbation Compositional Adversarial

Composite Attack

Papers

CBPF: Filtering Poisoned Data Based on Composite Backdoor Attack

A Survey on Vulnerability of Federated Learning: A Learning Algorithm Perspective

Prompt Packer: Deceiving LLMs through Compositional Instruction with Hidden Attacks

Composite Backdoor Attacks Against Large Language Models

Efficient Search of Comprehensively Robust Neural Architectures via Multi-fidelity Evaluation

Data-Driven Leader-following Consensus for Nonlinear Multi-Agent Systems against Composite Attacks: A Twins Layer Approach

Resilient Output Containment Control of Heterogeneous Multiagent Systems Against Composite Attacks: A Digital Twin Approach

Deep Composite Face Image Attacks: Generation, Vulnerability and Detection

A Closer Look at Robustness to L-infinity and Spatial Perturbations and their Composition

CARBEN: Composite Adversarial Robustness Benchmark

Towards Compositional Adversarial Robustness: Generalizing Adversarial Training to Composite Semantic Perturbations