Differential Privacy

Differential privacy (DP) is a rigorous framework for ensuring data privacy in machine learning by adding carefully calibrated noise to model training processes. Current research focuses on improving the accuracy of DP models, particularly for large-scale training, through techniques like adaptive noise allocation, Kalman filtering for noise reduction, and novel gradient processing methods. This active area of research is crucial for enabling the responsible use of sensitive data in various applications, ranging from healthcare and finance to natural language processing and smart grids, while maintaining strong privacy guarantees.

Papers

November 3, 2022

November 1, 2022

On the Interaction Between Differential Privacy and Gradient Compression in Deep Learning
Jimmy Lin
Deep Learning Differential Privacy Interaction Generation Gradient Compression Differential Privacy Guarantee Differential Privacy Mechanism

October 31, 2022

Fully Adaptive Composition for Gaussian Differential Privacy
Adam Smith, Abhradeep Thakurta
Differential Privacy Sensitive Data Gaussian Noise Privacy Budget

October 28, 2022

Differential Privacy has Bounded Impact on Fairness in Classification
Paul Mangold, Michaël Perrot, Aurélien Bellet, Marc Tommasi
Classification Code Differential Privacy Procedural Fairness Group Fairness Lipschitz Operator Private Model

October 27, 2022

Anonymized Histograms in Intermediate Privacy Models
Badih Ghazi, Pritish Kamath, Ravi Kumar, Pasin Manurangsi
Differential Privacy Histogram Layer Private Model Unlabeled Dataset Privacy Model Histogram Matching

October 26, 2022

October 25, 2022

October 24, 2022

October 22, 2022

October 20, 2022

Learning-Augmented Private Algorithms for Multiple Quantile Release
Mikhail Khodak, Kareem Amin, Travis Dick, Sergei Vassilvitskii
Differential Privacy Privacy Preserving Sensitive Data Quantile Regression Private Algorithm Learning Augmented Algorithm

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 18, 2022

October 17, 2022

Stochastic Differentially Private and Fair Learning
Andrew Lowy, Devansh Gupta, Meisam Razaviyayn
Differential Privacy Stochastic Way Fair Machine Learning Private Algorithm Fairness Notion Fair Learning

Differential Privacy

Papers

Distributed DP-Helmet: Scalable Differentially Private Non-interactive Averaging of Single Layers

Revisiting Hyperparameter Tuning with Differential Privacy

On the Interaction Between Differential Privacy and Gradient Compression in Deep Learning

Fully Adaptive Composition for Gaussian Differential Privacy

Differential Privacy has Bounded Impact on Fairness in Classification

Anonymized Histograms in Intermediate Privacy Models

Privately Fine-Tuning Large Language Models with Differential Privacy

Local Graph-homomorphic Processing for Privatized Distributed Systems

Synthetic Text Generation with Differential Privacy: A Simple and Practical Recipe

Streaming Submodular Maximization with Differential Privacy

Analyzing Privacy Leakage in Machine Learning via Multiple Hypothesis Testing: A Lesson From Fano

NVIDIA FLARE: Federated Learning from Simulation to Real-World

Generalised Likelihood Ratio Testing Adversaries through the Differential Privacy Lens

Federated Calibration and Evaluation of Binary Classifiers

Mixed Precision Quantization to Tackle Gradient Leakage Attacks in Federated Learning

Learning-Augmented Private Algorithms for Multiple Quantile Release

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

Differentially Private Diffusion Models

DPIS: An Enhanced Mechanism for Differentially Private SGD with Importance Sampling

Stochastic Differentially Private and Fair Learning