Issue |
ESAIM: PS
Volume 27, 2023
|
|
---|---|---|
Page(s) | 482 - 514 | |
DOI | https://doi.org/10.1051/ps/2023006 | |
Published online | 12 April 2023 |
Non-asymptotic analysis of Stochastic approximation algorithms for streaming data*
1
Sorbonne Université, Laboratoire de Probabilités, Statistique et Modélisation (LPSM), 75005 Paris, France
2
Wolfgang Pauli Institut, c/o Fakultät für Mathematik, Universität Wien, 1090 Vienna, Austria
** Corresponding author: nicklas.werge@sorbonne-universite.fr
Received:
16
March
2022
Accepted:
3
March
2023
We introduce a streaming framework for analyzing stochastic approximation/optimization problems. This streaming framework is analogous to solving optimization problems using time-varying mini-batches that arrive sequentially. We provide non-asymptotic convergence rates of various gradientbased algorithms; this includes the famous Stochastic Gradient (SG) descent (a.k.a. Robbins-Monro algorithm), mini-batch SG and time-varying mini-batch SG algorithms, as well as their iterated averages (a.k.a. Polyak-Ruppert averaging). We show (i) how to accelerate convergence by choosing the learning rate according to the time-varying mini-batches, (ii) that Polyak-Ruppert averaging achieves optimal convergence in terms of attaining the Cramer-Rao lower bound, and (iii) how time-varying mini-batches together with Polyak-Ruppert averaging can provide variance reduction and accelerate convergence simultaneously, which is advantageous for many learning problems, such as online, sequential, and large-scale learning. We further demonstrate these favorable effects for various time-varying minibatches.
Mathematics Subject Classification: 62L12 / 62L20 / 68W27 / 90C25
Key words: Stochastic algorithms / stochastic optimization / machine learning / online learning / mini-batch / streaming
© The authors. Published by EDP Sciences, SMAI 2023
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.