analiza aiqbench
This research advances performative reinforcement learning by establishing convergence guarantees for policy mixtures without relying on restrictive Lipschitz sensitivity assumptions, which often fail in multi-agent dynamics. The authors demonstrate that local and global stability are distinct concepts, providing a weighted Hedge dynamic that achieves local stationarity even with discontinuous environment maps while deriving unavoidable floors for global stability under trajectory feedback. These findings offer more robust theoretical foundations for deploying learning agents in environments where the policy actively shapes future data distributions.
Local and Global Stability in Performative Reinforcement Learning
arXiv:2609.06467v1 Announce Type: new Abstract: In performative reinforcement learning the deployed policy shapes the environment that generates the learner's future data, and the natural solution concept is a performatively stable policy that is optimal in the environment it induces. Existing convergence guarantees rely on Lipschitz sensitivity assumptions on the environment map $\pi \mapsto (P_\pi, r_\pi)$, which are hard to verify and fail in settings such as multi-agent best-response dynamics. We instead study stability for mixtures of policies, and show that the resulting picture is fundamentally different from performative prediction, where randomization removes the need for any sensitivity assumption. We distinguish local mixed stability, an occupancy-weighted first-order relaxation that we show is equivalent to stationarity, from global mixed stability, which certifies against arbitrary deviating policies. Our first result is that a weighted per-state Hedge dynamic drives the local stability gap to zero at an $O(1/\sqrt{T})$ rate for an arbitrary, possibly discontinuous, environment map, both with exact and with trajectory feedback. The two notions genuinely differ: we exhibit an instance where local stability is achieved exactly but every mixture has global stability gap bounded away from zero. For global stability we introduce a bounded transition range assumption, strictly weaker than Lipschitz sensitivity, under which unweighted per-state Hedge converges up to a floor of $O(\gamma\epsilon_P/(1-\gamma)^3)$, and we prove a matching-in-$\epsilon_P$ lower bound of $\Omega(\gamma\epsilon_P/(1-\gamma))$ under trajectory feedback, so this floor is unavoidable. Finally, we extend both notions to $n$-player performative Markov games, obtaining local stability with no assumption on the joint environment map or game structure, and global stability for performative Markov potential games.
Abstract:In performative reinforcement learning the deployed policy shapes the environment that generates the learner's future data, and the natural solution concept is a performatively stable policy that is optimal in the environment it induces. Existing convergence guarantees rely on Lipschitz sensitivity assumptions on the environment map $\pi \mapsto (P_\pi, r_\pi)$, which are hard to verify and fail in settings such as multi-agent best-response dynamics. We instead study stability for mixtures of policies, and show that the resulting picture is fundamentally different from performative prediction, where randomization removes the need for any sensitivity assumption. We distinguish local mixed stability, an occupancy-weighted first-order relaxation that we show is equivalent to stationarity, from global mixed stability, which certifies against arbitrary deviating policies. Our first result is that a weighted per-state Hedge dynamic drives the local stability gap to zero at an $O(1/\sqrt{T})$ rate for an arbitrary, possibly discontinuous, environment map, both with exact and with trajectory feedback. The two notions genuinely differ: we exhibit an instance where local stability is achieved exactly but every mixture has global stability gap bounded away from zero. For global stability we introduce a bounded transition range assumption, strictly weaker than Lipschitz sensitivity, under which unweighted per-state Hedge converges up to a floor of $O(\gamma\epsilon_P/(1-\gamma)^3)$, and we prove a matching-in-$\epsilon_P$ lower bound of $\Omega(\gamma\epsilon_P/(1-\gamma))$ under trajectory feedback, so this floor is unavoidable. Finally, we extend both notions to $n$-player performative Markov games, obtaining local stability with no assumption on the joint environment map or game structure, and global stability for performative Markov potential games.
Machine Learning (cs.LG)
arXiv:2609.06467 [cs.LG]
(or arXiv:2609.06467v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.06467 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 350 cuvinte
Subiecte conexe
News
Quantum Computing Report · 25 iul. 2026
Neuromorphic Reinforcement Learning for Quadruped Locomotion Control on Uneven Terrain
arXiv · 10 iun. 2026
Federated Learning for Feature Generalization with Convex Constraints
arXiv · 15 iun. 2026
StepOPSD: Step-Aware Online Preference Self-Distillation for Agent Reinforcement Learning
arXiv · 27 aug. 2026