Gradient Inversion

Gradient inversion is a technique used to reconstruct training data from model gradients, posing a significant privacy risk in federated learning and other distributed training settings. Current research focuses on improving the accuracy and efficiency of these attacks, particularly for high-resolution images and large batches, employing various architectures including deep neural networks and diffusion models, and exploring algorithms like neural architecture search and independent component analysis to enhance reconstruction quality. The ability to reconstruct sensitive data from gradients has major implications for data privacy in machine learning, driving the development of both improved attacks and robust defenses.

Papers

November 5, 2024

Gradient-Guided Conditional Diffusion Models for Private Image Reconstruction: Analyzing Adversarial Impacts of Differential Privacy and Denoising
Tao Huang, Jiayang Meng, Hong Chen, Guolong Zheng, Xu Yang, Xun Yi, Hua Wang
Differential Privacy Full State Reconstruction Conditional Diffusion Model Gradient Inversion Reconstructed Image Differential Privacy Noise Adversarial Influence Private Image

September 27, 2024

Gradient-free Decoder Inversion in Latent Diffusion Models
Seongmin Hong, Suh Yoon Jeon, Kyeonghyun Lee, Ernest K. Ryu, Se Young Chun
Latent Space Latent Diffusion Model Gradient Inversion

September 26, 2024

Federated Learning under Attack: Improving Gradient Inversion for Batch of Images
Luiz Leite, Yuri Santo, Bruno L. Dalmazo, André Riker
Attack Success Rate Gradient Inversion Attack Gradient Inversion Gradient Attack Batch Greenkhorn

September 18, 2024

InverseMeetInsert: Robust Real Image Editing via Geometric Accumulation Inversion in Guided Diffusion Models
Yan Zheng, Lemeng Wu
Diffusion Model Robust Version Image Editing Visual Prompting Gradient Inversion Real Image Editing

September 11, 2024

Exploring User-level Gradient Inversion with a Diffusion Prior
Zhuohang Li, Andrew Lowy, Jing Liu, Toshiaki Koike-Akino, Bradley Malin, Kieran Parsons, Ye Wang
Diffusion Model Diffusion Prior Gradient Inversion Attack Realistic Face Gradient Inversion Data Reconstruction

May 31, 2024

GI-NAS: Boosting Gradient Inversion Attacks through Adaptive Neural Architecture Search
Wenbo Yu, Hao Fang, Bin Chen, Xiaohang Sui, Chuan Chen, Hao Wu, Shu-Tao Xia, Ke Xu
Neural Architecture Search Prior Knowledge Gradient Inversion Attack Gradient Inversion

May 30, 2024

Gradient Inversion of Federated Diffusion Models
Jiyue Huang, Chi Hong, Lydia Y. Chen, Stefanie Roos
Diffusion Model Gradient Inversion Attack Gradient Inversion Inversion Attack Privacy Preserving Training

May 24, 2024

DAGER: Exact Gradient Inversion for Large Language Models
Ivo Petrov, Dimitar I. Dimitrov, Maximilian Baader, Mark Niklas Müller, Martin Vechev
Self Attention Layer Token Embeddings Gradient Inversion Attack Gradient Inversion Decoder Architecture

May 2, 2024

Recovering Labels from Local Updates in Federated Learning
Huancheng Chen, Haris Vikalo
Local Context Gradient Inversion Reconstructed Image Precise Inter EPOCH Feature Label Extraction Label Recovery

March 26, 2024

Leak and Learn: An Attacker's Cookbook to Train Using Leaked Data from Federated Learning
Joshua C. Zhao, Ahaan Dabholkar, Atul Sharma, Saurabh Bagchi
Private Training Gradient Inversion Data Reconstruction Attack Pseudo Victim AttackER Leakage Attack

March 17, 2024

CGI-DM: Digital Copyright Authentication for Diffusion Models via Contrasting Gradient Inversion
Xiaoyu Wu, Yang Hua, Chumeng Liang, Jiaru Zhang, Hao Wang, Tao Song, Haibing Guan
Diffusion Model Gradient Inversion Shot Natural Language Generation

March 13, 2024

March 6, 2024

SPEAR:Exact Gradient Inversion of Batches in Federated Learning
Dimitar I. Dimitrov, Maximilian Baader, Mark Niklas Müller, Martin Vechev
Gradient Inversion Attack Large Batch Low Rank Structure Gradient Inversion Gradient Sparsification Computational Reconstruction

December 19, 2023

Foreseeing Reconstruction Quality of Gradient Inversion: An Optimization Perspective
HyeongGwon Hong, Yooshin Cho, Hanbyel Cho, Jaesung Ahn, Junmo Kim
Gradient Norm Gradient Inversion Attack Gradient Inversion Reconstruction Quality

November 23, 2023

OASIS: Offsetting Active Reconstruction Attacks in Federated Learning
Tre' R. Jeter, Truc Nguyen, Raed Alharbi, My T. Thai
Attack Strategy Reconstruction Attack Gradient Inversion Step Attack

July 17, 2023

Domain Adaptation using Silver Standard Masks for Lateral Ventricle Segmentation in FLAIR MRI
Owen Crystal, Pejman J. Maralani, Sandra Black, Alan R. Moody, April Khademi
Domain Adaptation Gradient Inversion Ventricle Segmentation Gold Standard

May 31, 2023

Surrogate Model Extension (SME): A Fast and Accurate Weight Update Attack on Federated Learning
Junyi Zhu, Ruicong Yao, Matthew B. Blaschko
Surrogate Model Natural Gradient Gradient Flow Gradient Inversion Attack Gradient Inversion Small and Medium Sized Enterprise

October 19, 2022

Learning to Invert: Simple Adaptive Attacks for Gradient Inversion in Federated Learning
Ruihan Wu, Xiangyu Chen, Chuan Guo, Kilian Q. Weinberger
LeArning Abstract Differential Privacy Gradient Compression Gradient Inversion Attack Gradient Inversion Model Gradient

October 13, 2022

PDEBENCH: An Extensive Benchmark for Scientific Machine Learning
Makoto Takamoto, Timothy Praditia, Raphael Leiteritz, Dan MacKinlay, Francesco Alesiani, Dirk Pflüger, Mathias Niepert
Benchmark Study Scientific Machine Learning Gradient Inversion Classical Simulation

Gradient Inversion

Papers

Gradient-Guided Conditional Diffusion Models for Private Image Reconstruction: Analyzing Adversarial Impacts of Differential Privacy and Denoising

Gradient-free Decoder Inversion in Latent Diffusion Models

Federated Learning under Attack: Improving Gradient Inversion for Batch of Images

InverseMeetInsert: Robust Real Image Editing via Geometric Accumulation Inversion in Guided Diffusion Models

Exploring User-level Gradient Inversion with a Diffusion Prior

GI-NAS: Boosting Gradient Inversion Attacks through Adaptive Neural Architecture Search

Gradient Inversion of Federated Diffusion Models

DAGER: Exact Gradient Inversion for Large Language Models

Recovering Labels from Local Updates in Federated Learning

Leak and Learn: An Attacker's Cookbook to Train Using Leaked Data from Federated Learning

CGI-DM: Digital Copyright Authentication for Diffusion Models via Contrasting Gradient Inversion

RAF-GI: Towards Robust, Accurate and Fast-Convergent Gradient Inversion Attack in Federated Learning

MGIC: A Multi-Label Gradient Inversion Attack based on Canny Edge Detection on Federated Learning

SPEAR:Exact Gradient Inversion of Batches in Federated Learning

Foreseeing Reconstruction Quality of Gradient Inversion: An Optimization Perspective

OASIS: Offsetting Active Reconstruction Attacks in Federated Learning

Domain Adaptation using Silver Standard Masks for Lateral Ventricle Segmentation in FLAIR MRI

Surrogate Model Extension (SME): A Fast and Accurate Weight Update Attack on Federated Learning

Learning to Invert: Simple Adaptive Attacks for Gradient Inversion in Federated Learning

PDEBENCH: An Extensive Benchmark for Scientific Machine Learning