MrCoM: A Meta-Regularized World-Model Generalizing Across Multi-Scenarios
Published in AAAI 2026, 2026
In model-based RL, world models struggle to generalize across scenarios. MrCoM extracts scenario information through meta-state regularization and aligns model optimization with multi-scenario policy learning through meta-value regularization, yielding a world model that generalizes across scenarios.
Links: arXiv