Saved in:
Bibliographic Details
Main Authors: Gopi, Sivakanth, Lee, Yin Tat, Liu, Daogao, Shen, Ruoqi, Tian, Kevin
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2302.06085
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912783192293376
author Gopi, Sivakanth
Lee, Yin Tat
Liu, Daogao
Shen, Ruoqi
Tian, Kevin
author_facet Gopi, Sivakanth
Lee, Yin Tat
Liu, Daogao
Shen, Ruoqi
Tian, Kevin
contents The development of efficient sampling algorithms catering to non-Euclidean geometries has been a challenging endeavor, as discretization techniques which succeed in the Euclidean setting do not readily carry over to more general settings. We develop a non-Euclidean analog of the recent proximal sampler of [LST21], which naturally induces regularization by an object known as the log-Laplace transform (LLT) of a density. We prove new mathematical properties (with an algorithmic flavor) of the LLT, such as strong convexity-smoothness duality and an isoperimetric inequality, which are used to prove a mixing time on our proximal sampler matching [LST21] under a warm start. As our main application, we show our warm-started sampler improves the value oracle complexity of differentially private convex optimization in $\ell_p$ and Schatten-$p$ norms for $p \in [1, 2]$ to match the Euclidean setting [GLL22], while retaining state-of-the-art excess risk bounds [GLLST23]. We find our investigation of the LLT to be a promising proof-of-concept of its utility as a tool for designing samplers, and outline directions for future exploration.
format Preprint
id arxiv_https___arxiv_org_abs_2302_06085
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Algorithmic Aspects of the Log-Laplace Transform and a Non-Euclidean Proximal Sampler
Gopi, Sivakanth
Lee, Yin Tat
Liu, Daogao
Shen, Ruoqi
Tian, Kevin
Data Structures and Algorithms
Cryptography and Security
Machine Learning
Probability
Computation
The development of efficient sampling algorithms catering to non-Euclidean geometries has been a challenging endeavor, as discretization techniques which succeed in the Euclidean setting do not readily carry over to more general settings. We develop a non-Euclidean analog of the recent proximal sampler of [LST21], which naturally induces regularization by an object known as the log-Laplace transform (LLT) of a density. We prove new mathematical properties (with an algorithmic flavor) of the LLT, such as strong convexity-smoothness duality and an isoperimetric inequality, which are used to prove a mixing time on our proximal sampler matching [LST21] under a warm start. As our main application, we show our warm-started sampler improves the value oracle complexity of differentially private convex optimization in $\ell_p$ and Schatten-$p$ norms for $p \in [1, 2]$ to match the Euclidean setting [GLL22], while retaining state-of-the-art excess risk bounds [GLLST23]. We find our investigation of the LLT to be a promising proof-of-concept of its utility as a tool for designing samplers, and outline directions for future exploration.
title Algorithmic Aspects of the Log-Laplace Transform and a Non-Euclidean Proximal Sampler
topic Data Structures and Algorithms
Cryptography and Security
Machine Learning
Probability
Computation
url https://arxiv.org/abs/2302.06085