aalto1 untyped-item.component.html

Asynchronous stochastic Quasi-Newton MCMC for non-convex optimization supplementary document

Loading...
Thumbnail Image

Access rights

openAccess
publishedVersion

URL

Journal Title

Journal ISSN

Volume Title

A4 Artikkeli konferenssijulkaisussa

Major/Subject

Mcode

Degree programme

Language

en

Pages

8

Series

35th International Conference on Machine Learning, ICML 2018, Volume 11, pp. 4674-4683, Proceedings of Machine Learning Research ; Volume 80

Abstract

Recent studies have illustrated that stochastic gradient Markov Chain Monte Carlo techniques have a strong potential in non-convex optimization, where local and global convergence guarantees can be shown under certain conditions. By building up on this recent theory, in this study, we develop an asynchronous-parallel stochastic L-BFGS algorithm for non-convex optimization. The proposed algorithm is suitable for both distributed and shared-memory settings. We provide formal theoretical analysis and show that the proposed method achieves an ergodic convergence rate of {equation Presented} (N being the total number of iterations) and it can achieve a linear speedup under certain conditions. We perform several experiments on both synthetic and real datasets. The results support our theory and show that the proposed algorithm provides a significant speedup over the recently proposed synchronous distributed L-BFGS algorithm.

Description

Keywords

Other note

Citation

Simsekli, U, Yildiz, C, Nguyen, T H, Richard, G & Cemgil, A T 2018, Asynchronous stochastic Quasi-Newton MCMC for non-convex optimization supplementary document. in A Krause & J Dy (eds), 35th International Conference on Machine Learning, ICML 2018. vol. 11, Proceedings of Machine Learning Research, vol. 80, International Machine Learning Society, pp. 4674-4683, International Conference on Machine Learning, Stockholm, Sweden, 10/07/2018. < http://proceedings.mlr.press/v80/simsekli18a.html >

Endorsement

Review

Supplemented By

Referenced By