Research/paper/2026-09-21

New Reinforcement Learning Research on GRPO and Urban Network Control

Two independent research papers on reinforcement learning were published on arXiv. The first introduces $\lambda$-Controlled GRPO to manage flow-matching ratio instability, while the second presents deep meta-models and reinforcement policies for calibrating and controlling urban networks. Both studies contribute novel methodologies to the field of reinforcement learning.

What every outlet reports

  • Publication of two research papers on arXiv
  • One paper proposes $\lambda$-Controlled GRPO
  • The other paper explores RL for urban network control
  • Both papers are in reinforcement learning

All 2 articles

Related stories