Closing the Gap Between SVRG and TD-SVRG with Gradient Splitting

Arsenii Mustafin, Ioannis Paschalidis, Alex Olshevsky

Keywords: Temporal Difference learning, Reinforcement Learning, SVRG, Optimization

TL;DR: We prove a linear convergence time for an SVRG-inspired temporal difference method which is identical to the original convergence time bound of SVRG in the convex setting.

Abstract:

Temporal difference (TD) learning is a simple algorithm for policy evaluation
in reinforcement learning. The performance of TD learning is affected by high
variance and it can be naturally enhanced with variance reduction techniques, such
as the Stochastic Variance Reduced Gradient (SVRG) method. Recently, multiple
works have sought to fuse TD learning with SVRG to obtain a policy evaluation
method with a linear rate of convergence. However, the resulting convergence rate
is significantly weaker than what is achieved by SVRG in the setting of convex
optimization. In this work we utilize a recent interpretation of TD-learning as the
splitting of the gradient of an appropriately chosen function, thus simplifying the
algorithm and fusing TD with SVRG. We prove a linear convergence bound that
is identical to the convergence bound available for SVRG in the convex setting.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes

Please Choose The Closest Area That Your Submission Falls Into: Reinforcement Learning (eg, decision and control, planning, hierarchical RL, robotics)

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/arxiv:2211.16237/code)