Regularized Offline Policy Optimization with Posterior Hybrid Bayesian Belief

arXiv cs.AI Papers

Summary

This paper introduces Posterior Hybrid Bayesian Belief (PhyB), a framework that reformulates the expectation in Bayesian RL as a convex combination over dynamics models, enabling efficient regularized offline policy optimization with bounded objective discrepancy and state-of-the-art performance.

arXiv:2606.00680v1 Announce Type: new Abstract: Offline reinforcement learning (RL) aims to optimize policies from pre-collected datasets. A bottleneck of this paradigm is managing epistemic uncertainty, which arises from limited data coverage (sample-level) and the ambiguity in identifying transition dynamics from finite data (model-level). To provide a unified quantification of these uncertainties, Bayesian RL has been proposed by treating the dynamics model as a random variable and maintaining a corresponding belief. Despite its theoretical appeal, policy optimization in Bayesian RL remains computationally challenging as it requires solving composite objectives with expectations. Prior methods either employ search-based techniques with poor computational scalability or impose restrictive posterior assumptions that sacrifice the adaptability of Bayesian RL. To address these limitations, we propose Posterior Hybrid Bayesian Belief (PhyB), which reformulates the expectation as a convex combination over a subset of dynamics models. Theoretical analysis demonstrates that the objective discrepancy induced by this approximation remains bounded. Based on PhyB, we develop an iterative regularized policy optimization algorithm that provides metric-agnostic guarantees for monotonic improvement until convergence. Empirical results demonstrate that PhyB achieves state-of-the-art performance on various benchmarks.
Original Article
View Cached Full Text

Cached at: 06/02/26, 03:48 PM

# Regularized Offline Policy Optimization with Posterior Hybrid Bayesian Belief
Source: [https://arxiv.org/abs/2606.00680](https://arxiv.org/abs/2606.00680)
[View PDF](https://arxiv.org/pdf/2606.00680)

> Abstract:Offline reinforcement learning \(RL\) aims to optimize policies from pre\-collected datasets\. A bottleneck of this paradigm is managing epistemic uncertainty, which arises from limited data coverage \(sample\-level\) and the ambiguity in identifying transition dynamics from finite data \(model\-level\)\. To provide a unified quantification of these uncertainties, Bayesian RL has been proposed by treating the dynamics model as a random variable and maintaining a corresponding belief\. Despite its theoretical appeal, policy optimization in Bayesian RL remains computationally challenging as it requires solving composite objectives with expectations\. Prior methods either employ search\-based techniques with poor computational scalability or impose restrictive posterior assumptions that sacrifice the adaptability of Bayesian RL\. To address these limitations, we propose Posterior Hybrid Bayesian Belief \(PhyB\), which reformulates the expectation as a convex combination over a subset of dynamics models\. Theoretical analysis demonstrates that the objective discrepancy induced by this approximation remains bounded\. Based on PhyB, we develop an iterative regularized policy optimization algorithm that provides metric\-agnostic guarantees for monotonic improvement until convergence\. Empirical results demonstrate that PhyB achieves state\-of\-the\-art performance on various benchmarks\.

## Submission history

From: Hongqiang Lin \[[view email](https://arxiv.org/show-email/d0fd1c71/2606.00680)\] **\[v1\]**Sat, 30 May 2026 11:35:26 UTC \(3,720 KB\)

Similar Articles

Generative OOD-regularized Model-based Policy Optimization

arXiv cs.LG

Introduces GORMPO, a density-regularized offline RL algorithm that uses generative density modeling to restrict policy updates to high-density areas, achieving 17% improvement on a real-world medical dataset and outperforming state-of-the-art baselines.

Near-Future Policy Optimization

Hugging Face Daily Papers

Proposes Near-Future Policy Optimization (NPO), a mixed-policy RL method that accelerates convergence by learning from a later checkpoint of the same training run, boosting Qwen3-VL-8B-Instruct performance from 57.88 to 62.84.

Gradient Extrapolation-Based Policy Optimization

arXiv cs.LG

The article introduces Gradient Extrapolation-Based Policy Optimization (GXPO), a method that approximates multi-step lookahead in RL training for LLMs using only three backward passes. It demonstrates improved reasoning performance on math benchmarks over standard GRPO while maintaining fixed active-phase costs.