Feature Space Hijacking Attack

Feature space hijacking attacks exploit vulnerabilities in collaborative machine learning frameworks, such as split learning and federated learning, to manipulate model training and potentially reconstruct sensitive client data. Current research focuses on understanding the effectiveness of these attacks against various architectures, including deep neural networks, and exploring defense mechanisms like differential privacy and function secret sharing, though their efficacy remains a subject of ongoing investigation. The significance of this research lies in its implications for the security and privacy of sensitive data used in collaborative machine learning, impacting the development of trustworthy and robust AI systems across diverse applications.

Papers

October 27, 2024

Meta-Learning Approaches for Improving Detection of Unseen Speech Deepfakes
Ivan Kukanov, Janne Laakkonen, Tomi Kinnunen, Ville Hautamäki
Meta Learning Deep Fake Unseen Data Shot Adaptation Improved Detection Unseen Language Invisible Attack Feature Space Hijacking Attack

October 22, 2024

Evaluating the Effectiveness of Attack-Agnostic Features for Morphing Attack Detection
Laurent Colbois, Sébastien Marcel
Generative Adversarial Network Morphing Attack Attack Detection Generative Image Feature Space Hijacking Attack

October 10, 2024

Enhancing Federated Domain Adaptation with Multi-Domain Prototype-Based Federated Fine-Tuning
Jingyuan Zhang, Yiyang Duan, Shuaicheng Niu, Yang Cao, Wei Yang Bryan Lim
Fine Tuning Multi Domain Domain Performance Federated Domain Adaptation Feature Space Hijacking Attack

April 14, 2024

Make Split, not Hijack: Preventing Feature-Space Hijacking Attacks in Split Learning
Tanveer Khan, Mindaugas Budzys, Antonis Michalas
Split Learning PRIvacy Leakage Split and Fit Secret Sharing Encrypted Model Feature Space Hijacking Attack Hijacking Task

March 5, 2024

Federated Learning Under Attack: Exposing Vulnerabilities through Data Poisoning Attacks in Computer Networks
Ehsan Nowroozi, Imran Haider, Rahim Taheri, Mauro Conti
Security Vulnerability Data Poisoning Attack Computer Network Feature Space Hijacking Attack

October 16, 2023

DANAA: Towards transferable attacks with double adversarial neuron attribution
Zhibo Jin, Zhiyu Zhu, Xinyi Wang, Jiayu Zhang, Jun Shen, Huaming Chen
Feature Importance Hidden Layer Transferable Attack Feature Space Hijacking Attack Neuron Importance Attribution Attack

July 2, 2023

3D-IDS: Doubly Disentangled Dynamic Intrusion Detection
Chenyang Qiu, Yingsheng Geng, Junrui Lu, Kaida Chen, Shitong Zhu, Ya Su, Guoshun Nan, Can Zhang, Junsong Fu, Qimei Cui, Xiaofeng Tao
Network Intrusion Detection Anomaly Detection Performance Unknown Attack Feature Space Hijacking Attack

May 2, 2023

Boosting Adversarial Transferability via Fusing Logits of Top-1 Decomposed Feature
Juanjuan Weng, Zhiming Luo, Dazhen Lin, Shaozi Li, Zhun Zhong
Adversarial Attack Singular Value Decomposition Second Ranked Logits Adversarial Transferability Feature Decomposition Feature Space Hijacking Attack

April 19, 2023

Secure Split Learning against Property Inference, Data Reconstruction, and Feature Space Hijacking Attacks
Yunlong Mao, Zexi Xin, Zhenyu Li, Jue Hong, Qingyou Yang, Sheng Zhong
Neural Network Split Learning Split Computing Data Reconstruction Property Inference Feature Space Hijacking Attack

December 13, 2022

Object-fabrication Targeted Attack for Object Detection
Xuchong Zhang, Changfeng Sun, Haoliang Han, Hongbin Sun
Adversarial Attack Adversarial Example Image Attack Feature Space Hijacking Attack Object Generation Attack

March 31, 2022

Improving Adversarial Transferability via Neuron Attribution-Based Attacks
Jianping Zhang, Weibin Wu, Jen-tse Huang, Yizhan Huang, Wenxuan Wang, Yuxin Su, Michael R. Lyu
New Attack Single Neuron Level Transferable Adversarial Attack Feature Space Hijacking Attack Neuron Importance Attribution Attack

January 11, 2022

Feature Space Hijacking Attacks against Differentially Private Split Learning
Grzegorz Gawron, Philip Stubbings
Neural Network Differential Privacy Split Learning Input Reconstruction Feature Space Hijacking Attack

Feature Space Hijacking Attack

Papers

Meta-Learning Approaches for Improving Detection of Unseen Speech Deepfakes

Evaluating the Effectiveness of Attack-Agnostic Features for Morphing Attack Detection

Enhancing Federated Domain Adaptation with Multi-Domain Prototype-Based Federated Fine-Tuning

Make Split, not Hijack: Preventing Feature-Space Hijacking Attacks in Split Learning

Federated Learning Under Attack: Exposing Vulnerabilities through Data Poisoning Attacks in Computer Networks

DANAA: Towards transferable attacks with double adversarial neuron attribution

3D-IDS: Doubly Disentangled Dynamic Intrusion Detection

Boosting Adversarial Transferability via Fusing Logits of Top-1 Decomposed Feature

Secure Split Learning against Property Inference, Data Reconstruction, and Feature Space Hijacking Attacks

Object-fabrication Targeted Attack for Object Detection

Improving Adversarial Transferability via Neuron Attribution-Based Attacks

Feature Space Hijacking Attacks against Differentially Private Split Learning