Differential Privacy Noise

Differential privacy noise is added to data or model parameters during machine learning to protect individual privacy while preserving overall data utility. Current research focuses on mitigating the negative impact of this noise on model accuracy through techniques like low-pass filtering of gradients, post-training noise adjustment, and adaptive noise injection strategies tailored to specific algorithms (e.g., stochastic gradient descent, Bayesian methods). These advancements aim to improve the privacy-utility trade-off in various applications, including federated learning and distributed systems, by enabling more accurate models while maintaining strong privacy guarantees.

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

August 24, 2024

DOPPLER: Differentially Private Optimizers with Low-pass Filter for Privacy Noise Reduction
Xinwei Zhang, Zhiqi Bu, Mingyi Hong, Meisam Razaviyayn
Differential Privacy Private Stochastic Private Optimization Differential Privacy Noise Low Pas Doppler Velocity

June 27, 2024

Too Good to be True? Turn Any Model Differentially Private With DP-Weights
David Zagardo
Full Model Differential Privacy Private Stochastic Gradient Descent Dp SGD Differential Privacy Noise

May 4, 2024

PrivSGP-VR: Differentially Private Variance-Reduced Stochastic Gradient Push with Tight Utility Bounds
Zehan Zhu, Yan Huang, Xin Wang, Jinming Xu
Differential Privacy Decentralized Learning Private Stochastic Gradient Descent Dp SGD Differential Privacy Noise Gradient Variance

February 9, 2024

On the Privacy of Selection Mechanisms with Gaussian Noise
Jonathan Lebensold, Doina Precup, Borja Balle
Privacy Policy Gaussian Noise Differential Privacy Noise Evolutionary Mechanism Approximate Differential Privacy

January 2, 2024

PPBFL: A Privacy Protected Blockchain-based Federated Learning Model
Yang Li, Chunhe Xia, Wanshuang Lin, Tianbo Wang
Federated Learning Privacy Policy Blockchain Based Platform Differential Privacy Noise Adaptive Differential Privacy

July 1, 2023

Hiding in Plain Sight: Differential Privacy Noise Exploitation for Evasion-resilient Localized Poisoning Attacks in Multiagent Reinforcement Learning
Md Tamjid Hossain, Hung La
Differential Privacy Poisoning Attack Plain Sight Differential Privacy Noise Adversarial Estimator

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

April 4, 2023

A Communication-efficient Local Differentially Private Algorithm in Federated Optimization
Syed Eqbal Alam, Dhirendra Shukla, Shrisha Rao
Private Algorithm Communication Complexity Differential Privacy Guarantee Agent Communication Federated Optimization Differential Privacy Noise

February 24, 2023

Active Membership Inference Attack under Local Differential Privacy in Federated Learning
Truc Nguyen, Phung Lai, Khang Tran, NhatHai Phan, My T. Thai
Membership Inference Attack Local Differential Privacy Membership Inference Private Training Differential Privacy Noise

January 31, 2023

Differentially Private Distributed Bayesian Linear Regression with MCMC
Barış Alparslan, Sinan Yıldırım, Ş. İlker Birbil
Markov Chain Monte Carlo Private Linear Regression Bayesian Estimation Bayesian Regression Differential Privacy Noise Bayesian Inference Framework

January 19, 2023

Differentially Private Online Bayesian Estimation With Adaptive Truncation
Sinan Yıldırım
Sequential Monte Carlo Differential Privacy Noise Online Bayesian Private Bayesian

November 14, 2022

A Robust Dynamic Average Consensus Algorithm that Ensures both Differential Privacy and Accurate Convergence
Yongqiang Wang
Differential Privacy Average Consensus Differential Privacy Noise

October 22, 2022

Mixed Precision Quantization to Tackle Gradient Leakage Attacks in Federated Learning
Pretom Roy Ovi, Emon Dey, Nirmalya Roy, Aryya Gangopadhyay
Differential Privacy Mixed Precision Quantization Gradient Leakage Attack Differential Privacy Noise Privacy Inference

August 18, 2022

Private, Efficient, and Accurate: Protecting Models Trained by Multi-party Learning with Differential Privacy
Wenqiang Ruan, Mingxin Xu, Wenjing Fang, Li Wang, Lei Wang, Weili Han
High Efficiency Differential Privacy Multi Party Private Text Party Computation Model Protection Differential Privacy Noise Cryptographic Protocol

Differential Privacy Noise

Papers

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

DOPPLER: Differentially Private Optimizers with Low-pass Filter for Privacy Noise Reduction

Too Good to be True? Turn Any Model Differentially Private With DP-Weights

PrivSGP-VR: Differentially Private Variance-Reduced Stochastic Gradient Push with Tight Utility Bounds

On the Privacy of Selection Mechanisms with Gaussian Noise

PPBFL: A Privacy Protected Blockchain-based Federated Learning Model

Hiding in Plain Sight: Differential Privacy Noise Exploitation for Evasion-resilient Localized Poisoning Attacks in Multiagent Reinforcement Learning

Securing Distributed SGD against Gradient Leakage Threats

A Communication-efficient Local Differentially Private Algorithm in Federated Optimization

Active Membership Inference Attack under Local Differential Privacy in Federated Learning

Differentially Private Distributed Bayesian Linear Regression with MCMC

Differentially Private Online Bayesian Estimation With Adaptive Truncation

A Robust Dynamic Average Consensus Algorithm that Ensures both Differential Privacy and Accurate Convergence

Mixed Precision Quantization to Tackle Gradient Leakage Attacks in Federated Learning

Private, Efficient, and Accurate: Protecting Models Trained by Multi-party Learning with Differential Privacy