notesum.ai

Published at October 22

Episodic Future Thinking Mechanism for Multi-agent Reinforcement Learning

cs.LG

cs.AI

Released Date: October 22, 2024

Authors: Dongsu Lee¹, Minhae Kwon²

Aff.: ¹Department of Intelligent Semiconductors, Soongsil University, Seoul, South Korea; ²School of Electronic Engineering, Soongsil University, Seoul, South Korea

Arxiv: https://arxiv.org/abs/2410.17373v1

Algorithm	The number of character groups ( $n$ )
Algorithm	$1$	$2$	$3$	$4$	$5$
Proposed	$2899$	$\mathbf{3047}$	$\mathbf{2976}$	$\mathbf{2948}$	$\mathbf{3051}$
FCE-EFT	$2899$	$2784$	$2646$	$2566$	$2629$
MADDPG [29]	$2763$	$\mathbf{3006}$	$2800$	$\mathbf{2933}$	$2856$
MAPPO [63]	$2753$	$2862$	$2597$	$2529$	$2763$
Q-MIX [44]	$2199$	$2310$	$2288$	$2118$	$1861$
Dreamer [15]	$2911$	$2813$	$2733$	$2631$	$2701$
MBPO [16]	$2089$	$1964$	$1523$	$1893$	$1633$
ToMC2 [59]	$\mathbf{3016}$	$2812$	$2683$	$2691$	$2511$
LIAM [38]	$1913$	$1792$	$1771$	$1683$	$1733$