Oracles & Followers: Stackelberg Equilibria in Deep Multi-Agent Reinforcement Learning
This work addresses the challenge of efficient equilibrium computation in strategic settings like security games, offering a flexible solution for researchers and practitioners, though it builds incrementally on existing methods.
The paper tackles the problem of finding Stackelberg equilibria in multi-agent reinforcement learning by proposing a general framework that accommodates various algorithmic designs, and introduces a contextual policy approach that demonstrates significantly improved sample efficiency in benchmark domains.
Stackelberg equilibria arise naturally in a range of popular learning problems, such as in security games or indirect mechanism design, and have received increasing attention in the reinforcement learning literature. We present a general framework for implementing Stackelberg equilibria search as a multi-agent RL problem, allowing a wide range of algorithmic design choices. We discuss how previous approaches can be seen as specific instantiations of this framework. As a key insight, we note that the design space allows for approaches not previously seen in the literature, for instance by leveraging multitask and meta-RL techniques for follower convergence. We propose one such approach using contextual policies, and evaluate it experimentally on both standard and novel benchmark domains, showing greatly improved sample efficiency compared to previous approaches. Finally, we explore the effect of adopting algorithm designs outside the borders of our framework.