Gradient Obfuscation

Gradient obfuscation encompasses techniques designed to hinder the recovery of sensitive information from gradients used in machine learning, particularly within federated learning and adversarial defense contexts. Current research focuses on developing both offensive (attacks exploiting obfuscated gradients) and defensive methods (e.g., modifying model architectures or employing novel optimization algorithms) to improve privacy and robustness against adversarial examples. This area is crucial for advancing privacy-preserving machine learning and enhancing the security of deployed models, impacting fields ranging from healthcare to finance.

Papers

July 17, 2024

Private and Federated Stochastic Convex Optimization: Efficient Strategies for Centralized Systems
Roie Reshef, Kfir Y. Levy
Differential Privacy Stochastic Optimization Stochastic Convex Optimization Simple Strategy Gradient Obfuscation Federated Automatic Differentiation

December 7, 2023

Privacy-preserving quantum federated learning via gradient hiding
Changhao Li, Niraj Kumar, Zhixin Song, Shouvanik Chakrabarti, Marco Pistoia
LeArning Abstract Quantum Machine Learning Quantum Computing Quantum Physic Quantum Circuit Quantum Channel Gradient Obfuscation Quantum Protocol

October 27, 2023

DiffAttack: Evasion Attacks Against Diffusion-Based Adversarial Purification
Mintong Kang, Dawn Song, Bo Li
Adversarial Example Evasion Attack Diffusion Purification Gradient Obfuscation

July 25, 2023

On the Leakage of Fuzzy Matchers
Axel Durbet, Kevin Thiry-Atighehchi, Dorine Chagnon, Paul-Marie Grollemund
Data Leakage Side Channel Deep Leakage Gradient Obfuscation Fuzzy String Matching

April 28, 2023

NNSplitter: An Active Defense Solution for DNN Model via Automated Weight Obfuscation
Tong Zhou, Yukui Luo, Shaolei Ren, Xiaolin Xu
Deep Neural Network Intellectual Property Protection Model Protection Gradient Obfuscation Active Defense

August 1, 2022

Attacking Adversarial Defences by Smoothing the Loss Landscape
Panagiotis Eustratiadis, Henry Gouk, Da Li, Timothy Hospedales
Adversarial Attack Loss Landscape Adversarial DEfense Gradient Based Adversarial Gradient Obfuscation

June 8, 2022

Gradient Obfuscation Gives a False Sense of Security in Federated Learning
Kai Yue, Richeng Jin, Chau-Wai Wong, Dror Baron, Huaiyu Dai
Security Related Gradient Compression Privacy Preserving Machine Learning False Sense Data Reconstruction Attack Gradient Obfuscation

June 3, 2022

Gradient Obfuscation Checklist Test Gives a False Sense of Security
Nikola Popovic, Danda Pani Paudel, Thomas Probst, Luc Van Gool
Adversarial Attack Security Related False Sense Gradient Obfuscation Stochastic Defense

May 21, 2022

Gradient Concealment: Free Lunch for Defending Adversarial Attacks
Sen Pei, Jiaxi Sun, Xiaopeng Zhang, Gaofeng Meng
Deep Learning Adversarial Attack Adversarial Example Adversarial Input Free Lunch Gradient Obfuscation

May 8, 2022

Decentralized Stochastic Optimization with Inherent Privacy Protection
Yongqiang Wang, H. Vincent Poor
Differential Privacy Decentralized Optimization Inherent Privacy Gradient Obfuscation Decentralized Stochastic Decentralized Stochastic Optimization

March 23, 2022

Enhancing Classifier Conservativeness and Robustness by Polynomiality
Ziqi Wang, Marco Loog
Adversarial Attack Native Robustness Softmax Function Gradient Obfuscation