Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Hu, Xiaobo, Lin, Youfang, Liu, Yue, Wang, Jinwen, Wang, Shuo, Fan, Hehe, Lv, Kai
Format: Preprint
Veröffentlicht: 2023
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2312.01915
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915487682658304
author Hu, Xiaobo
Lin, Youfang
Liu, Yue
Wang, Jinwen
Wang, Shuo
Fan, Hehe
Lv, Kai
author_facet Hu, Xiaobo
Lin, Youfang
Liu, Yue
Wang, Jinwen
Wang, Shuo
Fan, Hehe
Lv, Kai
contents Visual reinforcement learning has proven effective in solving control tasks with high-dimensional observations. However, extracting reliable and generalizable representations from vision-based observations remains a central challenge. Inspired by the human thought process, when the representation extracted from the observation can predict the future and trace history, the representation is reliable and accurate in comprehending the environment. Based on this concept, we introduce a Bidirectional Transition (BiT) model, which leverages the ability to bidirectionally predict environmental transitions both forward and backward to extract reliable representations. Our model demonstrates competitive generalization performance and sample efficiency on two settings of the DeepMind Control suite. Additionally, we utilize robotic manipulation and CARLA simulators to demonstrate the wide applicability of our method.
format Preprint
id arxiv_https___arxiv_org_abs_2312_01915
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Learning Robust Representations via Bidirectional Transition for Visual Reinforcement Learning
Hu, Xiaobo
Lin, Youfang
Liu, Yue
Wang, Jinwen
Wang, Shuo
Fan, Hehe
Lv, Kai
Computer Vision and Pattern Recognition
Visual reinforcement learning has proven effective in solving control tasks with high-dimensional observations. However, extracting reliable and generalizable representations from vision-based observations remains a central challenge. Inspired by the human thought process, when the representation extracted from the observation can predict the future and trace history, the representation is reliable and accurate in comprehending the environment. Based on this concept, we introduce a Bidirectional Transition (BiT) model, which leverages the ability to bidirectionally predict environmental transitions both forward and backward to extract reliable representations. Our model demonstrates competitive generalization performance and sample efficiency on two settings of the DeepMind Control suite. Additionally, we utilize robotic manipulation and CARLA simulators to demonstrate the wide applicability of our method.
title Learning Robust Representations via Bidirectional Transition for Visual Reinforcement Learning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2312.01915