# Convergence of Online Adaptive and Recurrent Optimization Algorithms

Abstract : We prove local convergence of several notable gradient descent algorithms used in machine learning, for which standard stochastic gradient descent theory does not apply directly. This includes, first, online algorithms for recurrent models and dynamical systems, such as \emph{Real-time recurrent learning} (RTRL) and its computationally lighter approximations NoBackTrack and UORO; second, several adaptive algorithms such as RMSProp, online natural gradient, and Adam with $\beta^2\to 1$. Despite local convergence being a relatively weak requirement for a new optimization algorithm, no local analysis was available for these algorithms, as far as we knew. Analysis of these algorithms does not immediately follow from standard stochastic gradient (SGD) theory. In fact, Adam has been proved to lack local convergence in some simple situations \citep{j.2018on}. For recurrent models, online algorithms modify the parameter while the model is running, which further complicates the analysis with respect to simple SGD. Local convergence for these various algorithms results from a single, more general set of assumptions, in the setup of learning dynamical systems online. Thus, these results can cover other variants of the algorithms considered. We adopt an ergodic'' rather than probabilistic viewpoint, working with empirical time averages instead of probability distributions. This is more data-agnostic and creates differences with respect to standard SGD theory, especially for the range of possible learning rates. For instance, with cycling or per-epoch reshuffling over a finite dataset instead of pure i.i.d.\ sampling with replacement, empirical averages of gradients converge at rate $1/T$ instead of $1/\sqrt{T}$ (cycling acts as a variance reduction method), theoretically allowing for larger learning rates than in SGD.
Document type :
Preprints, Working Papers, ...
Domain :

https://hal.archives-ouvertes.fr/hal-02566660
Contributor : Pierre-Yves Massé Connect in order to contact the contributor
Submitted on : Wednesday, December 16, 2020 - 4:58:48 PM
Last modification on : Tuesday, January 19, 2021 - 10:16:03 AM

### Files

main.pdf
Files produced by the author(s)

### Identifiers

• HAL Id : hal-02566660, version 2
• ARXIV : 2005.05645

### Citation

Pierre-Yves Massé, Yann Ollivier. Convergence of Online Adaptive and Recurrent Optimization Algorithms. 2020. ⟨hal-02566660v2⟩

Record views