Private Sequential Learning
Name
1805.02136.pdf
Description
Accepted version
Size
583.63 KB
Format
Adobe PDF
Checksum (MD5)
8ee1f44fd3109184d57f515e42870e4e
Author(s)
Tsitsiklis, John N
Xu, Kuang
Xu, Zhi
Date Issued
2021
Journal
Operations Research
Publisher
Institute for Operations Research and the Management Sciences (INFORMS)
Citation
Tsitsiklis, John N, Xu, Kuang and Xu, Zhi. 2021. "Private Sequential Learning." Operations Research, 69 (5).
Version
Author's final manuscript
Abstract
Can we learn privately and efficiently through sequential interactions? A private learning model is formulated to study an intrinsic tradeoff between privacy and query complexity in sequential learning. The formulation involves a learner who aims to learn a scalar value by sequentially querying an external database and receiving binary responses. In the meantime, an adversary observes the learner’s queries, although not the responses, and tries to infer from them the scalar value of interest. The objective of the learner is to obtain an accurate estimate of the scalar value using only a small number of queries while simultaneously protecting his or her privacy by making the scalar value provably difficult to learn for the adversary. The main results provide tight upper and lower bounds on the learner’s query complexity as a function of desired levels of privacy and estimation accuracy. The authors also construct explicit query strategies whose complexity is optimal up to an additive constant.
MIT Department
Massachusetts Institute of Technology. Laboratory for Information and Decision Systems
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1287/OPRE.2020.2021
