notesum.ai

Published at November 10

A Variance Minimization Approach to Temporal-Difference Learning

cs.LG

cs.AI

Released Date: November 10, 2024

Authors: Xingguo Chen¹, Yu Gong, Shangdong Yang¹, Wenhao Wang²

Aff.: ¹Jiangsu Key Laboratory of Big Data Security and Intelligent Processing, Nanjing University of Posts and Telecommunications, Nanjing 210023, China; ²College of Electronic Engineering, National University of Defense Technology, China

Arxiv: http://arxiv.org/abs/2411.06396v1

algorithm	TD	VMTD	TDC	VMTDC	ETD	VMETD
ON-POLICY	$\bm{0.475}$	$0.25$	$\bm{0.09025}$	$0.025$	$\bm{4.75}$	$2.5$
OFF-POLICY	$-0.2$	$\bm{0.25}$	$0.016$	$\bm{0.025}$	$\bm{3.4}$	$1.15$