Data-Augmented Game Starts for Accelerating Self-Play Exploration in Imperfect Information Games
For researchers working on large-scale imperfect-information games, this method addresses the exploration bottleneck caused by sparse rewards and long horizons, offering a practical way to leverage offline data to speed up equilibrium finding.
The paper proposes Data-Augmented Game Starts (DAGS), a multi-agent starting-state sampling strategy that uses offline human demonstrations to initialize reinforcement learning data collection at intermediate states, accelerating exploration in regularized policy-gradient methods for two-player zero-sum games. Experiments show DAGS achieves lower exploitability under fixed computational budgets in games with challenging exploration.
Finding approximate equilibria for large-scale imperfect-information competitive games such as StarCraft, Dota, and CounterStrike remains computationally infeasible due to sparse rewards and challenging exploration over long horizons. In this paper, we propose a multi-agent starting-state sampling strategy designed to substantially accelerate online exploration in regularized policy-gradient game methods for two-player zero-sum (2p0s) games. Motivated by an assumption that offline demonstrations from skilled humans can provide good coverage of high-level strategies relevant to equilibrium play, we propose the initialization of reinforcement learning data collection at intermediate states sampled from offline data to facilitate exploration of strategically relevant subgames. Referring to this method as Data-Augmented Game Starts (DAGS), we perform experiments using synthetic datasets and analytically tractable, long-horizon control variants of two-player Kuhn Poker, Goofspiel, and a counterexample game designed to penalize biased beliefs over hidden information. Under fixed computational budgets, DAGS enables regularized policy gradient methods to achieve lower exploitability in games with significantly more challenging exploration. We show that augmenting starting state distributions when solving imperfect information games can lead to biased equilibria, and we provide a straightforward mitigation to this in the form of multi-task observation flags. Finally, we release a new set of benchmark environments that drastically increase exploration challenges and state counts in existing OpenSpiel games while keeping exploitability measurements analytically tractable.