Saved in:
Bibliographic Details
Main Authors: Lee, Jongmin, Sun, Meiqi, Abbeel, Pieter
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.10042
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911312606396416
author Lee, Jongmin
Sun, Meiqi
Abbeel, Pieter
author_facet Lee, Jongmin
Sun, Meiqi
Abbeel, Pieter
contents In the unsupervised pre-training for reinforcement learning, the agent aims to learn a prior policy for downstream tasks without relying on task-specific reward functions. We focus on state entropy maximization (SEM), where the goal is to learn a policy that maximizes the entropy of the state stationary distribution. In this paper, we introduce SEMDICE, a principled off-policy algorithm that computes an SEM policy from an arbitrary off-policy dataset, which optimizes the policy directly within the space of stationary distributions. SEMDICE computes a single, stationary Markov state-entropy-maximizing policy from an arbitrary off-policy dataset. Experimental results demonstrate that SEMDICE outperforms baseline algorithms in maximizing state entropy while achieving the best adaptation efficiency for downstream tasks among SEM-based unsupervised RL pre-training methods.
format Preprint
id arxiv_https___arxiv_org_abs_2512_10042
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SEMDICE: Off-policy State Entropy Maximization via Stationary Distribution Correction Estimation
Lee, Jongmin
Sun, Meiqi
Abbeel, Pieter
Machine Learning
In the unsupervised pre-training for reinforcement learning, the agent aims to learn a prior policy for downstream tasks without relying on task-specific reward functions. We focus on state entropy maximization (SEM), where the goal is to learn a policy that maximizes the entropy of the state stationary distribution. In this paper, we introduce SEMDICE, a principled off-policy algorithm that computes an SEM policy from an arbitrary off-policy dataset, which optimizes the policy directly within the space of stationary distributions. SEMDICE computes a single, stationary Markov state-entropy-maximizing policy from an arbitrary off-policy dataset. Experimental results demonstrate that SEMDICE outperforms baseline algorithms in maximizing state entropy while achieving the best adaptation efficiency for downstream tasks among SEM-based unsupervised RL pre-training methods.
title SEMDICE: Off-policy State Entropy Maximization via Stationary Distribution Correction Estimation
topic Machine Learning
url https://arxiv.org/abs/2512.10042