SkyNet extends MuZero to partially-observable stochastic games by adding auxiliary belief-aware heads, significantly outperforming baselines in complex card games.
March 31, 2026
Original Paper
SkyNet: Belief-Aware Planning for Partially-Observable Stochastic Games
arXiv · 2603.27751
The Takeaway
Addresses a major limitation of model-based RL in environments with hidden information (like finance or negotiation). It proves that latent states can be forced to capture belief distributions without explicit state tracking or search algorithm changes.
From the abstract
In 2019, Google DeepMind released MuZero, a model-based reinforcement learning method that achieves strong results in perfect-information games by combining learned dynamics models with Monte Carlo Tree Search (MCTS). However, comparatively little work has extended MuZero to partially observable, stochastic, multi-player environments, where agents must act under uncertainty about hidden state. Such settings arise not only in card games but in domains such as autonomous negotiation, financial tra