AI Paper Atlas

Research map / Reinforcement learning / MDPs and planning

Finite Sample Analysis of Average-Reward TD Learning and $Q$-Learning

Sheng Zhang, Zhe Zhang, Siva Theja Maguluri

NeurIPS 2021 · Reinforcement learning › MDPs and planning · approach: Function approximation in RL theory · 6 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
vary lambda in {0, 0.2, 0.4, 0.8} for TD(lambda) with diminishing step sizes (ablation)Average-reward TD(lambda) with linear function approximation · Random MRP, 100 states, 20 featuressample complexity / error decayMixed
start the algorithm from 4 different initial theta_0Average-reward TD(lambda) · Random MRP, 100 states, 20 featuresE||Pi_{2,E}(theta_t - theta*)||_2No clear effect
start the algorithm from 4 different initial theta_0Average-reward TD(lambda) · Random MRP, 100 states, 20 featurescomponent of (theta_t - theta*) along theta_eMixed
use diminishing step sizes beta_t = c1/(t+c2) instead of a constant step sizeAverage-reward TD(lambda) · Theorem 1mean-square distance to TD fixed setImproved
add an intermediate lambda rather than very small or very largeAverage-reward TD(lambda) sample complexity bound · Corollary 1sample complexityImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy