Pith. sign in

REVIEW 3 cited by

A Lyapunov Theory for Finite-Sample Guarantees of Asynchronous Q-Learning and TD-Learning Variants

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2102.01567 v4 pith:QKDDNVFS submitted 2021-02-02 cs.LG math.OCstat.ML

classification cs.LGmath.OCstat.ML
keywords algorithmsconvergenceasynchronousboundsfinite-samplefirstguaranteeslambda
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

This paper develops an unified framework to study finite-sample convergence guarantees of a large class of value-based asynchronous reinforcement learning (RL) algorithms. We do this by first reformulating the RL algorithms as \textit{Markovian Stochastic Approximation} (SA) algorithms to solve fixed-point equations. We then develop a Lyapunov analysis and derive mean-square error bounds on the convergence of the Markovian SA. Based on this result, we establish finite-sample mean-square convergence bounds for asynchronous RL algorithms such as $Q$-learning, $n$-step TD, TD$(\lambda)$, and off-policy TD algorithms including V-trace. As a by-product, by analyzing the convergence bounds of $n$-step TD and TD$(\lambda)$, we provide theoretical insights into the bias-variance trade-off, i.e., efficiency of bootstrapping in RL. This was first posed as an open problem in (Sutton, 1999).

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Finite-Time Analysis of Discounted Exponential-Utility Reinforcement Learning

    cs.LG 2026-08 accept novelty 7.0 of 10

    The one- and two-timescale algorithms for discounted exponential-utility RL achieve O~(1/sqrt(n)) finite-time rates under Markovian sampling with parameter-free stepsizes.

  2. Sharp asymptotic theory for Q-learning with LDTZ learning rate and its generalization

    stat.ML 2026-04 unverdicted novelty 6.0 of 10

    Q-learning with PD2Z/LD2Z step sizes admits sharp non-asymptotic bounds, a tail Polyak–Ruppert CLT, and a time-uniform Gaussian approximation, establishing a best-of-both-worlds rate-and-bias tradeoff.

  3. Statistical and Algorithmic Foundations of Reinforcement Learning

    stat.ML 2025-07 accept

    A tutorial collecting minimax sample complexity results for tabular RL across generative model, online, offline, robust, and human-feedback settings.

Pith tools