public-data assisted private stochastic optimization
Public-data Assisted Private Stochastic Optimization: Power and Limitations
We study the limits and capability of public-data assisted differentially private (PA-DP) algorithms. Specifically, we focus on the problem of stochastic convex optimization (SCO) with either labeled or unlabeled public data. These lower bounds are established via our new lower bounds for PA-DP mean estimation, which are of a similar form. Up to constant factors, these lower bounds show that the simple strategy of either treating all data as private or discarding the private data, is optimal. We also study PA-DP supervised learning with \textit{unlabeled} public samples.
artificial intelligence, machine learning, public-data assisted private stochastic optimization, (11 more...)
Technology: Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (0.40)