Private optimization in the interpolation regime: faster rates and hardness results

Hilal Asi, Karan Chadha, Gary Cheng, John Duchi
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:1025-1045, 2022.

Abstract

In non-private stochastic convex optimization, stochastic gradient methods converge much faster on interpolation problems—namely, problems where there exists a solution that simultaneously minimizes all of the sample losses—than on non-interpolating ones; similar improvements are not known in the private setting. In this paper, we investigate differentially private stochastic optimization in the interpolation regime. First, we show that without additional assumptions, interpolation problems do not exhibit an improved convergence rates with differential privacy. However, when the functions exhibit quadratic growth around the optimum, we show (near) exponential improvements in the private sample complexity. In particular, we propose an adaptive algorithm that improves the sample complexity to achieve expected error $\alpha$ from $\frac{d}{\diffp \sqrt{\alpha}}$ to $\frac{1}{\alpha^\rho} + \frac{d}{\diffp} \log\paren{\frac{1}{\alpha}}$ for any fixed $\rho >0$, while retaining the standard minimax-optimal sample complexity for non-interpolation problems. We prove a lower bound that shows the dimension-dependent term in the expression above is tight. Furthermore, we provide a superefficiency result which demonstrates the necessity of the polynomial term for adaptive algorithms: any algorithm that has a polylogarithmic sample complexity for interpolation problems cannot achieve the minimax-optimal rates for the family of non-interpolation problems.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-asi22a, title = {Private optimization in the interpolation regime: faster rates and hardness results}, author = {Asi, Hilal and Chadha, Karan and Cheng, Gary and Duchi, John}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {1025--1045}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/asi22a/asi22a.pdf}, url = {https://proceedings.mlr.press/v162/asi22a.html}, abstract = {In non-private stochastic convex optimization, stochastic gradient methods converge much faster on interpolation problems—namely, problems where there exists a solution that simultaneously minimizes all of the sample losses—than on non-interpolating ones; similar improvements are not known in the private setting. In this paper, we investigate differentially private stochastic optimization in the interpolation regime. First, we show that without additional assumptions, interpolation problems do not exhibit an improved convergence rates with differential privacy. However, when the functions exhibit quadratic growth around the optimum, we show (near) exponential improvements in the private sample complexity. In particular, we propose an adaptive algorithm that improves the sample complexity to achieve expected error $\alpha$ from $\frac{d}{\diffp \sqrt{\alpha}}$ to $\frac{1}{\alpha^\rho} + \frac{d}{\diffp} \log\paren{\frac{1}{\alpha}}$ for any fixed $\rho >0$, while retaining the standard minimax-optimal sample complexity for non-interpolation problems. We prove a lower bound that shows the dimension-dependent term in the expression above is tight. Furthermore, we provide a superefficiency result which demonstrates the necessity of the polynomial term for adaptive algorithms: any algorithm that has a polylogarithmic sample complexity for interpolation problems cannot achieve the minimax-optimal rates for the family of non-interpolation problems.} }
Endnote
%0 Conference Paper %T Private optimization in the interpolation regime: faster rates and hardness results %A Hilal Asi %A Karan Chadha %A Gary Cheng %A John Duchi %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-asi22a %I PMLR %P 1025--1045 %U https://proceedings.mlr.press/v162/asi22a.html %V 162 %X In non-private stochastic convex optimization, stochastic gradient methods converge much faster on interpolation problems—namely, problems where there exists a solution that simultaneously minimizes all of the sample losses—than on non-interpolating ones; similar improvements are not known in the private setting. In this paper, we investigate differentially private stochastic optimization in the interpolation regime. First, we show that without additional assumptions, interpolation problems do not exhibit an improved convergence rates with differential privacy. However, when the functions exhibit quadratic growth around the optimum, we show (near) exponential improvements in the private sample complexity. In particular, we propose an adaptive algorithm that improves the sample complexity to achieve expected error $\alpha$ from $\frac{d}{\diffp \sqrt{\alpha}}$ to $\frac{1}{\alpha^\rho} + \frac{d}{\diffp} \log\paren{\frac{1}{\alpha}}$ for any fixed $\rho >0$, while retaining the standard minimax-optimal sample complexity for non-interpolation problems. We prove a lower bound that shows the dimension-dependent term in the expression above is tight. Furthermore, we provide a superefficiency result which demonstrates the necessity of the polynomial term for adaptive algorithms: any algorithm that has a polylogarithmic sample complexity for interpolation problems cannot achieve the minimax-optimal rates for the family of non-interpolation problems.
APA
Asi, H., Chadha, K., Cheng, G. & Duchi, J.. (2022). Private optimization in the interpolation regime: faster rates and hardness results. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:1025-1045 Available from https://proceedings.mlr.press/v162/asi22a.html.

Related Material