Tsukada Lab. Computer Network Group

FM-EAC: Feature Model-based Enhanced Actor-Critic for Multi-Task Control in Dynamic Environments

Quanxi Zhou, Wencan Mao, Manabu Tsukada, John C. S. Lui, Yusheng Ji

, Dec 17, 2025

Summary in three lines

  • FM-EAC addresses generalization across tasks and changing environments in reinforcement learning.
  • A feature model extracts information about the current scenario for both actor and critic.
  • The dynamic-environment simulations report higher rewards than the baseline actor–critic.

Figures from the paper

Figure 1: FM-EAC framework
Figure 1. FM-EAC frameworkSource: Figure 1 from Zhou et al., arXiv (2025) (authors’ figure from the manuscript)The feature model supplies a representation of the environment to actor and critic.

How to cite

Q. Zhou, W. Mao, M. Tsukada, J. C. S. Lui, and Y. Ji, “FM-EAC: Feature Model-based Enhanced Actor-Critic for Multi-Task Control in Dynamic Environments,” in , 2025.

BibTeX

@unpublished{Zhou2025a,
  title         = {FM-EAC: Feature Model-based Enhanced Actor-Critic for Multi-Task Control in Dynamic Environments},
  author        = {Quanxi Zhou and Wencan Mao and Manabu Tsukada and John C. S. Lui and Yusheng Ji},
  year          = {2025},
  date          = {2025-12-17},
  eprint        = {2512.15430},
  archiveprefix = {arXiv},
  primaryclass  = {cs.LG},
  url           = {https://arxiv.org/abs/2512.15430},
  note          = {Preprint}
}