Saved in:
Bibliographic Details
Main Authors: Kazemi, Milad, Perez, Mateo, Somenzi, Fabio, Soudjani, Sadegh, Trivedi, Ashutosh, Velasquez, Alvaro
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.15693
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914409486483456
author Kazemi, Milad
Perez, Mateo
Somenzi, Fabio
Soudjani, Sadegh
Trivedi, Ashutosh
Velasquez, Alvaro
author_facet Kazemi, Milad
Perez, Mateo
Somenzi, Fabio
Soudjani, Sadegh
Trivedi, Ashutosh
Velasquez, Alvaro
contents Recent advances in reinforcement learning (RL) have renewed interest in reward design for shaping agent behavior, but manually crafting reward functions is tedious and error-prone. A principled alternative is to specify behavioral requirements in a formal, unambiguous language and automatically compile them into learning objectives. $ω$-regular languages are a natural fit, given their role in formal verification and synthesis. However, most existing $ω$-regular RL approaches operate in an episodic, discounted setting with periodic resets, which is misaligned with $ω$-regular semantics over infinite traces. For continuing tasks, where the agent interacts with the environment over a single uninterrupted lifetime, the average-reward criterion is more appropriate. We focus on absolute liveness specifications, a subclass of $ω$-regular languages that cannot be violated by any finite prefix and thus aligns naturally with continuing interaction. We present the first model-free RL framework that translates absolute liveness specifications into average-reward objectives and enables learning in unknown communicating Markov decision processes (MDPs) without episodic resetting. We also introduce a reward structure for lexicographic multi-objective optimization: among policies that maximize the satisfaction probability of an absolute liveness specification, the agent maximizes an external average-reward objective. Our method guarantees convergence in unknown communicating MDPs and supports on-the-fly reductions that do not require full environment knowledge, enabling model-free learning. Experiments across several benchmarks show that the continuing, average-reward approach outperforms competing discount-based methods.
format Preprint
id arxiv_https___arxiv_org_abs_2505_15693
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Average Reward Reinforcement Learning for Omega-Regular and Mean-Payoff Objectives
Kazemi, Milad
Perez, Mateo
Somenzi, Fabio
Soudjani, Sadegh
Trivedi, Ashutosh
Velasquez, Alvaro
Artificial Intelligence
Recent advances in reinforcement learning (RL) have renewed interest in reward design for shaping agent behavior, but manually crafting reward functions is tedious and error-prone. A principled alternative is to specify behavioral requirements in a formal, unambiguous language and automatically compile them into learning objectives. $ω$-regular languages are a natural fit, given their role in formal verification and synthesis. However, most existing $ω$-regular RL approaches operate in an episodic, discounted setting with periodic resets, which is misaligned with $ω$-regular semantics over infinite traces. For continuing tasks, where the agent interacts with the environment over a single uninterrupted lifetime, the average-reward criterion is more appropriate. We focus on absolute liveness specifications, a subclass of $ω$-regular languages that cannot be violated by any finite prefix and thus aligns naturally with continuing interaction. We present the first model-free RL framework that translates absolute liveness specifications into average-reward objectives and enables learning in unknown communicating Markov decision processes (MDPs) without episodic resetting. We also introduce a reward structure for lexicographic multi-objective optimization: among policies that maximize the satisfaction probability of an absolute liveness specification, the agent maximizes an external average-reward objective. Our method guarantees convergence in unknown communicating MDPs and supports on-the-fly reductions that do not require full environment knowledge, enabling model-free learning. Experiments across several benchmarks show that the continuing, average-reward approach outperforms competing discount-based methods.
title Average Reward Reinforcement Learning for Omega-Regular and Mean-Payoff Objectives
topic Artificial Intelligence
url https://arxiv.org/abs/2505.15693