Gradient Leakage

Gradient leakage refers to the vulnerability of machine learning models, particularly in federated learning settings, where sensitive training data can be reconstructed from shared gradient updates. Current research focuses on developing and analyzing both attacks that exploit this vulnerability (often leveraging diffusion models or gradient inversion techniques) and defenses, including noise injection (e.g., differential privacy), gradient compression, and architectural modifications (e.g., variational bottlenecks). Understanding and mitigating gradient leakage is crucial for ensuring data privacy in distributed machine learning applications, impacting the security and trustworthiness of various systems, from image processing to network intrusion detection.

Papers

November 5, 2024

FEDLAD: Federated Evaluation of Deep Leakage Attacks and Defenses
Isaac Baglin, Xiatian Zhu, Simon Hadfield
Federated Prompt Cooperation Leakage Attack Gradient Leakage

October 11, 2024

Gradients Stand-in for Defending Deep Leakage in Federated Learning
H. Yi, H. Ren, C. Hu, Y. Li, J. Deng, X. Xie
Sensitive Data Gradient Leakage Model Gradient Aggregated Gradient Deep Leakage

July 7, 2024

Gradient Diffusion: A Perturbation-Resilient Gradient Leakage Attack
Xuan Liu, Siqi Cai, Qihua Zhou, Song Guo, Ruibin Li, Kaiwei Lin
Gradient Leakage Gradient Leakage Attack Gradient Perturbation Gradient Guided Diffusion

June 13, 2024

Is Diffusion Model Safe? Severe Data Leakage via Gradient-Guided Diffusion Model
Jiayang Meng, Tao Huang, Hong Chen, Cuiping Li
Differential Privacy Image Reconstruction Reconstruction Attack Gradient Leakage Information Leakage Gradient Guided Diffusion

April 8, 2024

SoK: Gradient Leakage in Federated Learning
Jiacheng Du, Jiahui Hu, Zhibo Wang, Peng Sun, Neil Zhenqiang Gong, Kui Ren
Gradient Inversion Attack Defense Method Private Training Gradient Leakage

February 2, 2024

Privacy-Preserving Distributed Learning for Residential Short-Term Load Forecasting
Yi Dong, Yingjie Wang, Mariana Gama, Mustafa A. Mustafa, Geert Deconinck, Xiaowei Huang
LeArning Abstract Energy Consumption Load Forecasting Secure Aggregation Gradient Leakage

November 22, 2023

A Theoretical Insight into Attack and Defense of Gradient Leakage in Transformer
Chenyang Li, Zhao Song, Weixin Wang, Chiwun Yang
Transformer Based Differential Privacy Theoretical Analysis Gradient Leakage Gradient Attack Gradient Leakage Attack

September 22, 2023

Understanding Deep Gradient Leakage via Inversion Influence Functions
Haobo Zhang, Junyuan Hong, Yuyang Deng, Mehrdad Mahdavi, Jiayu Zhou
PRIvacy Leakage Inversion Performance Private Gradient Gradient Leakage

September 8, 2023

Privacy Preserving Federated Learning with Convolutional Variational Bottlenecks
Daniel Scheliga, Patrick Mäder, Marco Seeland
Gradient Inversion Attack Privacy Preserving Federated Learning Gradient Leakage Gradient Leakage Attack Variational Bottleneck

May 10, 2023

Securing Distributed SGD against Gradient Leakage Threats
Wenqi Wei, Ling Liu, Jingya Zhou, Ka-Ho Chow, Yanzhao Wu
Differential Privacy Federated Learning Stochastic Gradient Descent Gradient Leakage Gradient Leakage Attack Differential Privacy Noise

May 6, 2023

Gradient Leakage Defense with Key-Lock Module for Federated Learning
Hanchi Ren, Jingjing Deng, Xianghua Xie, Xiaoke Ma, Jianfeng Ma
Privacy Preserving Machine Learning Gradient Leakage Gradient Leakage Attack Door Status

November 14, 2022

Impact of spiking neurons leakages and network recurrences on event-based spatio-temporal pattern recognition
Mohamed Sadek Bouanane, Dalila Cherifi, Elisabetta Chicca, Lyes Khacef
Global Impact Neuromorphic Hardware Neuron Model Energy Efficient Inference Event Based Object Recognition Gradient Leakage Energy Efficient Neuromorphic Recurrence Network

October 15, 2022

Sketching for First Order Method: Efficient Algorithm for Low-Bandwidth Channel and Vulnerability
Zhao Song, Yitan Wang, Zheng Yu, Lichen Zhang
Efficient Algorithm Hand Drawn Sketch Latent Vulnerability Gradient Information Gradient Based Attack Gradient Leakage First Order Method Sketching System Low Bitrates

October 8, 2022

FedDef: Defense Against Gradient Leakage in Federated Learning-based Network Intrusion Detection Systems
Jiahui Chen, Yi Zhao, Qi Li, Xuewei Feng, Ke Xu
Adversarial Attack Federated Learning Gradient Leakage Robust Federated Learning Adversarial Traffic

September 13, 2022

Concealing Sensitive Samples against Gradient Leakage in Federated Learning
Jing Wu, Munawar Hayat, Mingyi Zhou, Mehrtash Harandi
Sensitive Data Private Data Model Inversion Attack Gradient Information Gradient Leakage

August 12, 2022

Dropout is NOT All You Need to Prevent Gradient Leakage
Daniel Scheliga, Patrick Mäder, Marco Seeland
Structured Dropout Gradient Inversion Attack Gradient Leakage Inversion Attack

August 9, 2022

Combining Variational Modeling with Partial Gradient Perturbation to Prevent Deep Gradient Leakage
Daniel Scheliga, Patrick Mäder, Marco Seeland
Gradient Inversion Attack Variational Model Gradient Leakage Gradient Attack Gradient Perturbation

May 26, 2022

Encoded Gradients Aggregation against Gradient Leakage in Federated Learning
Dun Zeng, Shiyu Liu, Siqi Liang, Zonghang Li, Hui Wang, Irwin King, Zenglin Xu
Homomorphic Encryption Secure Aggregation Gradient Leakage Aggregated Gradient

March 29, 2022

Auditing Privacy Defenses in Federated Learning via Generative Gradient Leakage
Zhuohang Li, Jiaxin Zhang, Luyang Liu, Jian Liu
Generative Adversarial Network Gradient Information Empirical Privacy Private Training Gradient Leakage

December 28, 2021

APRIL: Finding the Achilles' Heel on Privacy for Vision Transformers
Jiahao Lu, Xi Sheryl Zhang, Tianli Zhao, Xiangyu He, Jian Cheng
Vision Transformer Privacy Policy PRIvacy Leakage Transformer Inference Gradient Leakage Gradient Leakage Attack