PrefPoE: Advantage-Guided Preference Fusion for Learning Where to Explore

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lin, Zhihao, Wu, Lin, Tian, Zhen, Lan, Jianglin
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911259478196224
author Lin, Zhihao
Wu, Lin
Tian, Zhen
Lan, Jianglin
author_facet Lin, Zhihao
Wu, Lin
Tian, Zhen
Lan, Jianglin
contents Exploration in reinforcement learning remains a critical challenge, as naive entropy maximization often results in high variance and inefficient policy updates. We introduce \textbf{PrefPoE}, a novel \textit{Preference-Product-of-Experts} framework that performs intelligent, advantage-guided exploration via the first principled application of product-of-experts (PoE) fusion for single-task exploration-exploitation balancing. By training a preference network to concentrate probability mass on high-advantage actions and fusing it with the main policy through PoE, PrefPoE creates a \textbf{soft trust region} that stabilizes policy updates while maintaining targeted exploration. Across diverse control tasks spanning both continuous and discrete action spaces, PrefPoE demonstrates consistent improvements: +321\% on HalfCheetah-v4 (1276~$\rightarrow$~5375), +69\% on Ant-v4, +276\% on LunarLander-v2, with consistently enhanced training stability and sample efficiency. Unlike standard PPO, which suffers from entropy collapse, PrefPoE sustains adaptive exploration through its unique dynamics, thereby preventing premature convergence and enabling superior performance. Our results establish that learning \textit{where to explore} through advantage-guided preferences is as crucial as learning how to act, offering a general framework for enhancing policy gradient methods across the full spectrum of reinforcement learning domains. Code and pretrained models are available in supplementary materials.
format Preprint
id arxiv_https___arxiv_org_abs_2511_08241
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PrefPoE: Advantage-Guided Preference Fusion for Learning Where to Explore
Lin, Zhihao
Wu, Lin
Tian, Zhen
Lan, Jianglin
Machine Learning
Exploration in reinforcement learning remains a critical challenge, as naive entropy maximization often results in high variance and inefficient policy updates. We introduce \textbf{PrefPoE}, a novel \textit{Preference-Product-of-Experts} framework that performs intelligent, advantage-guided exploration via the first principled application of product-of-experts (PoE) fusion for single-task exploration-exploitation balancing. By training a preference network to concentrate probability mass on high-advantage actions and fusing it with the main policy through PoE, PrefPoE creates a \textbf{soft trust region} that stabilizes policy updates while maintaining targeted exploration. Across diverse control tasks spanning both continuous and discrete action spaces, PrefPoE demonstrates consistent improvements: +321\% on HalfCheetah-v4 (1276~$\rightarrow$~5375), +69\% on Ant-v4, +276\% on LunarLander-v2, with consistently enhanced training stability and sample efficiency. Unlike standard PPO, which suffers from entropy collapse, PrefPoE sustains adaptive exploration through its unique dynamics, thereby preventing premature convergence and enabling superior performance. Our results establish that learning \textit{where to explore} through advantage-guided preferences is as crucial as learning how to act, offering a general framework for enhancing policy gradient methods across the full spectrum of reinforcement learning domains. Code and pretrained models are available in supplementary materials.
title PrefPoE: Advantage-Guided Preference Fusion for Learning Where to Explore
topic Machine Learning
url https://arxiv.org/abs/2511.08241