# Policy Iteration for Two-Player General-Sum Stochastic Stackelberg Games

arXiv: https://arxiv.org/abs/2405.06689 · HTML: https://mlsanity.com/paper/2405.06689/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 22 references that are in the map.
- First posted on arXiv 7 May 2024.

## Builds on (most cited first, 8 of 8)

- [Vulnerability of Deep Reinforcement Learning to Policy Induction Attacks](https://mlsanity.com/paper/1701.04143/) (2017)
- [Adaptive Reward-Poisoning Attacks against Reinforcement Learning](https://mlsanity.com/paper/2003.12613/) (2020)
- [Policy Teaching via Environment Poisoning: Training-time Adversarial Attacks against Reinforcement Learning](https://mlsanity.com/paper/2003.12909/) (2020)
- [Deceptive Reinforcement Learning Under Adversarial Manipulations on Cost Signals](https://mlsanity.com/paper/1906.10571/) (2019)
- [Bi-level Actor-Critic for Multi-agent Coordination](https://mlsanity.com/paper/1909.03510/) (2019)
- [Vulnerability-Aware Poisoning Mechanism for Online RL with Unknown Dynamics](https://mlsanity.com/paper/2009.00774/) (2020)
- [Online Learning in Stackelberg Games with an Omniscient Follower](https://mlsanity.com/paper/2301.11518/) (2023)
- [Decentralized Cooperative Reinforcement Learning with Hierarchical Information Structure](https://mlsanity.com/paper/2111.00781/) (2021)

## Abstract (arXiv)

We address two-player general-sum stochastic Stackelberg games (SSGs), where the leader's policy is optimized considering the best-response follower whose policy is optimal for its reward under the leader. Existing policy gradient and value iteration approaches for SSGs do not guarantee monotone improvement in the leader's policy under the best-response follower. Consequently, their performance is not guaranteed when their limits are not stationary Stackelberg equilibria (SSEs), which do not necessarily exist. In this paper, we derive a policy improvement theorem for SSGs under the best-response follower and propose a novel policy iteration algorithm that guarantees monotone improvement in the leader's performance. Additionally, we introduce Pareto-optimality as an extended optimality of the SSE and prove that our method converges to the Pareto front when the leader is myopic.