$κ$-Explorer: A Unified Framework for Active Model Estimation in MDPs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Gu, Xihe, Mitra, Urbashi, Javidi, Tara
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911464245166080
author Gu, Xihe
Mitra, Urbashi
Javidi, Tara
author_facet Gu, Xihe
Mitra, Urbashi
Javidi, Tara
contents In tabular Markov decision processes (MDPs) with perfect state observability, each trajectory provides active samples from the transition distributions conditioned on state-action pairs. Consequently, accurate model estimation depends on how the exploration policy allocates visitation frequencies in accordance with the intrinsic complexity of each transition distribution. Building on recent work on coverage-based exploration, we introduce a parameterized family of decomposable and concave objective functions $U_κ$ that explicitly incorporate both intrinsic estimation complexity and extrinsic visitation frequency. Moreover, the curvature $κ$ provides a unified treatment of various global objectives, such as the average-case and worst-case estimation error objectives. Using the closed-form characterization of the gradient of $U_κ$, we propose $κ$-Explorer, an active exploration algorithm that performs Frank-Wolfe-style optimization over state-action occupancy measures. The diminishing-returns structure of $U_κ$ naturally prioritizes underexplored and high-variance transitions, while preserving smoothness properties that enable efficient optimization. We establish tight regret guarantees for $κ$-Explorer and further introduce a fully online and computationally efficient surrogate algorithm for practical use. Experiments on benchmark MDPs demonstrate that $κ$-Explorer provides superior performance compared to existing exploration strategies.
format Preprint
id arxiv_https___arxiv_org_abs_2602_20404
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle $κ$-Explorer: A Unified Framework for Active Model Estimation in MDPs
Gu, Xihe
Mitra, Urbashi
Javidi, Tara
Machine Learning
In tabular Markov decision processes (MDPs) with perfect state observability, each trajectory provides active samples from the transition distributions conditioned on state-action pairs. Consequently, accurate model estimation depends on how the exploration policy allocates visitation frequencies in accordance with the intrinsic complexity of each transition distribution. Building on recent work on coverage-based exploration, we introduce a parameterized family of decomposable and concave objective functions $U_κ$ that explicitly incorporate both intrinsic estimation complexity and extrinsic visitation frequency. Moreover, the curvature $κ$ provides a unified treatment of various global objectives, such as the average-case and worst-case estimation error objectives. Using the closed-form characterization of the gradient of $U_κ$, we propose $κ$-Explorer, an active exploration algorithm that performs Frank-Wolfe-style optimization over state-action occupancy measures. The diminishing-returns structure of $U_κ$ naturally prioritizes underexplored and high-variance transitions, while preserving smoothness properties that enable efficient optimization. We establish tight regret guarantees for $κ$-Explorer and further introduce a fully online and computationally efficient surrogate algorithm for practical use. Experiments on benchmark MDPs demonstrate that $κ$-Explorer provides superior performance compared to existing exploration strategies.
title $κ$-Explorer: A Unified Framework for Active Model Estimation in MDPs
topic Machine Learning
url https://arxiv.org/abs/2602.20404