Enregistré dans:
Détails bibliographiques
Auteurs principaux: Wu, Qingyuan, Wang, Yuhui, Zhan, Simon Sinong, Wang, Yixuan, Lin, Chung-Wei, Lv, Chen, Zhu, Qi, Schmidhuber, Jürgen, Huang, Chao
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2505.00546
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866918049523695616
author Wu, Qingyuan
Wang, Yuhui
Zhan, Simon Sinong
Wang, Yixuan
Lin, Chung-Wei
Lv, Chen
Zhu, Qi
Schmidhuber, Jürgen
Huang, Chao
author_facet Wu, Qingyuan
Wang, Yuhui
Zhan, Simon Sinong
Wang, Yixuan
Lin, Chung-Wei
Lv, Chen
Zhu, Qi
Schmidhuber, Jürgen
Huang, Chao
contents Reinforcement learning (RL) with delays is challenging as sensory perceptions lag behind the actual events: the RL agent needs to estimate the real state of its environment based on past observations. State-of-the-art (SOTA) methods typically employ recursive, step-by-step forecasting of states. This can cause the accumulation of compounding errors. To tackle this problem, our novel belief estimation method, named Directly Forecasting Belief Transformer (DFBT), directly forecasts states from observations without incrementally estimating intermediate states step-by-step. We theoretically demonstrate that DFBT greatly reduces compounding errors of existing recursively forecasting methods, yielding stronger performance guarantees. In experiments with D4RL offline datasets, DFBT reduces compounding errors with remarkable prediction accuracy. DFBT's capability to forecast state sequences also facilitates multi-step bootstrapping, thus greatly improving learning efficiency. On the MuJoCo benchmark, our DFBT-based method substantially outperforms SOTA baselines. Code is available at https://github.com/QingyuanWuNothing/DFBT.
format Preprint
id arxiv_https___arxiv_org_abs_2505_00546
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Directly Forecasting Belief for Reinforcement Learning with Delays
Wu, Qingyuan
Wang, Yuhui
Zhan, Simon Sinong
Wang, Yixuan
Lin, Chung-Wei
Lv, Chen
Zhu, Qi
Schmidhuber, Jürgen
Huang, Chao
Machine Learning
Reinforcement learning (RL) with delays is challenging as sensory perceptions lag behind the actual events: the RL agent needs to estimate the real state of its environment based on past observations. State-of-the-art (SOTA) methods typically employ recursive, step-by-step forecasting of states. This can cause the accumulation of compounding errors. To tackle this problem, our novel belief estimation method, named Directly Forecasting Belief Transformer (DFBT), directly forecasts states from observations without incrementally estimating intermediate states step-by-step. We theoretically demonstrate that DFBT greatly reduces compounding errors of existing recursively forecasting methods, yielding stronger performance guarantees. In experiments with D4RL offline datasets, DFBT reduces compounding errors with remarkable prediction accuracy. DFBT's capability to forecast state sequences also facilitates multi-step bootstrapping, thus greatly improving learning efficiency. On the MuJoCo benchmark, our DFBT-based method substantially outperforms SOTA baselines. Code is available at https://github.com/QingyuanWuNothing/DFBT.
title Directly Forecasting Belief for Reinforcement Learning with Delays
topic Machine Learning
url https://arxiv.org/abs/2505.00546