New RaCO-DP method enhances private learning with fairness constraints

Researchers have developed RaCO-DP, a novel method for optimizing machine learning models under differential privacy while adhering to rate constraints. This approach addresses challenges in applying standard DP techniques to objectives that depend on aggregate statistics across subpopulations, such as group fairness constraints. RaCO-DP utilizes a Lagrangian formulation and a novel analysis of Stochastic Gradient Descent-Ascent (SGDA) to ensure privacy and convergence. Empirical results indicate that RaCO-DP outperforms existing private learning methods in terms of privacy, utility, and fairness, particularly for neural networks. AI

IMPACT Enhances privacy and fairness in machine learning models, potentially improving trustworthy AI applications.

RANK_REASON The cluster contains a research paper detailing a new method for private rate-constrained optimization in machine learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• differential privacy
• DP SGD
• Hugging Face
• IArxiv
• RaCO-DP
• ScienceCast
• sgdA
• Tudor Cebere

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →