Bit Flip Attack

Bit-flip attacks exploit vulnerabilities in the memory storing neural network parameters, causing subtle alterations that significantly degrade model accuracy or even introduce malicious behavior. Current research focuses on developing both attack techniques tailored to specific architectures like graph neural networks and convolutional neural networks, and defense mechanisms ranging from encoding schemes to in-DRAM protection and runtime integrity verification. This area is crucial for securing deployed deep learning models, particularly in safety-critical applications, as even single-bit flips can have catastrophic consequences. The ongoing development of both more sophisticated attacks and robust defenses highlights the critical need for enhanced security in machine learning systems.

Papers

November 12, 2024

Impactful Bit-Flip Search on Full-precision Models
Nadav Benedek, Matan Levy, Mahmood Sharif
Search Query Latent Vulnerability Accurate Model Bit Flip Attack Bit Flip

May 22, 2024

DeepNcode: Encoding-Based Protection against Bit-Flip Attacks on Neural Networks
Patrik Velčický, Jakub Breier, Mladen Kovačević, Xiaolu Hou
Neural Network New Attack Bit Flip Attack Fault Injection Attack

April 24, 2024

An Empirical Study of Aegis
Daniel Saragih, Paridhi Goel, Tejas Balaji, Alyssa Li
Adversarial Attack Native Robustness Adversarial Example Empirical Study Bit Flip Attack Training Robustness

November 2, 2023

Attacking Graph Neural Networks with Bit Flips: Weisfeiler and Lehman Go Indifferent
Lorenz Kummer, Samir Moustafa, Nils N. Kriege, Wilfried N. Gansterer
Graph Neural Network Weisfeiler Leman Bit Flip Attack Bit Flip Flip Attack Fault Injection Attack

October 5, 2023

Burning the Adversarial Bridges: Robust Windows Malware Detection Against Binary-level Mutations
Ahmed Abusnaina, Yizhen Wang, Sunpreet Arora, Ke Wang, Mihai Christodorescu, David Mohaisen
Malware Detection Adversarial Malware Bit Flip Attack

September 12, 2023

Compiled Models, Built-In Exploits: Uncovering Pervasive Bit-Flip Attack Surfaces in DNN Executables
Yanzuo Chen (1), Zhibo Liu (1), Yuanyuan Yuan (1), Sihang Hu (2), Tianxiang Li (2), Shuai Wang (1) ((1) The Hong Kong University of Science and Technology, (2) Huawei Technologies)
DNN Framework Bit Flip Attack Bit Flip

August 12, 2023

One-bit Flip is All You Need: When Bit-flip Attack Meets Model Training
Jianshuo Dong, Han Qiu, Yiming Li, Tianwei Zhang, Yuanjie Li, Zeqi Lai, Chao Zhang, Shu-Tao Xia
Model Training Malicious Model Bit Flip Attack Bit Flip Flip Attack Bit Training

May 14, 2023

DNN-Defender: A Victim-Focused In-DRAM Defense Mechanism for Taming Adversarial Weight Attack on DNNs
Ranyang Zhou, Sabbir Ahmed, Adnan Siraj Rakin, Shaahin Angizi
Deep Neural Network DEfense Mechanism Bit Flip Attack 1T1C DRAM Cell

December 7, 2021

BDFA: A Blind Data Adversarial Bit-flip Attack on Deep Neural Networks
Behnam Ghavami, Mani Sadati, Mohammad Shahidzadeh, Zhenman Fang, Lesley Shannon
Deep Neural Network Feedback Alignment Accuracy Loss Bit Flip Attack Bit Flip Flip Attack

November 2, 2021

HASHTAG: Hash Signatures for Online Detection of Fault-Injection Attacks on Deep Neural Networks
Mojan Javaheripi, Farinaz Koushanfar
Deep Neural Network Automatic Detection DNN Model Digital Signature Bit Flip Attack Fault Injection Attack

Bit Flip Attack

Papers

Impactful Bit-Flip Search on Full-precision Models

DeepNcode: Encoding-Based Protection against Bit-Flip Attacks on Neural Networks

An Empirical Study of Aegis

Attacking Graph Neural Networks with Bit Flips: Weisfeiler and Lehman Go Indifferent

Burning the Adversarial Bridges: Robust Windows Malware Detection Against Binary-level Mutations

Compiled Models, Built-In Exploits: Uncovering Pervasive Bit-Flip Attack Surfaces in DNN Executables

One-bit Flip is All You Need: When Bit-flip Attack Meets Model Training

DNN-Defender: A Victim-Focused In-DRAM Defense Mechanism for Taming Adversarial Weight Attack on DNNs

BDFA: A Blind Data Adversarial Bit-flip Attack on Deep Neural Networks

HASHTAG: Hash Signatures for Online Detection of Fault-Injection Attacks on Deep Neural Networks