Entity · technique

Private Stochastic Convex Optimization

techniqueactiveprivate-stochastic-convex-optimization-669f1dd4·1 events·first seen May 18, 2026

Aliases: Private Stochastic Convex Optimization

Co-occurring entities

Differential Privacy Approximate DP CVaR (Conditional Value at Risk)

More like this (12)

Accelerated Decentralized Stochastic Gradient Descent for Strongly Convex Optimization Proximal Policy Optimization differentiable convex optimization Bayesian Optimization distributionally robust optimization distributed optimization Biconvex Optimization Pareto Optimal Policy Optimization Divergence Regularized Policy Optimization Denoising Diffusion Policy Optimization stochastic gradient ascent Bayesian Multiobjective Optimization

Recent events (1)

4arXiv · cs.LG·May 18, 2026·source ↗

The Privacy Price of Tail-Risk Learning: Effective Tail Sample Size in Differentially Private CVaR Optimization

This paper characterizes how differential privacy affects the statistical complexity of CVaR (Conditional Value at Risk) optimization, showing that the effective sample size governing private tail-risk learning is εnτ rather than n, where τ is the tail mass. Complete minimax rates are derived for scalar estimation and finite classes under pure DP, with lower bounds extending to approximate DP. For convex Lipschitz learning, the CVaR-specific privacy cost necessarily scales as 1/(εnτ), with dimension dependence inherited from private stochastic convex optimization. The results reduce private CVaR learning to private learning on Θ(nτ) tail records as the canonical hard subproblem.

AI Safety Research Differential Privacy Approximate DP Private Stochastic Convex Optimization +1 more