Unsupervised Reinforcement Learning

Unsupervised reinforcement learning (URL) aims to train agents to learn effective behaviors without relying on explicit reward signals, focusing instead on intrinsic motivations like exploration and skill discovery. Current research emphasizes developing algorithms that efficiently explore state spaces, often employing techniques like empowerment, entropy maximization/minimization, and curiosity-driven exploration, sometimes within model-based or ensemble frameworks. These advancements are significant because they promise more data-efficient and generalizable agents, potentially impacting robotics, personalized medicine (e.g., EEG analysis), and other fields requiring adaptable, self-improving systems.

Papers

October 2, 2022

EUCLID: Towards Efficient Unsupervised Reinforcement Learning with Multi-choice Dynamics Model
Yifu Yuan, Jianye Hao, Fei Ni, Yao Mu, Yan Zheng, Yujing Hu, Jinyi Liu, Yingfeng Chen, Changjie Fan
Unsupervised Pre Training Unsupervised Reinforcement Learning Transition Dynamic

September 24, 2022

Mastering the Unsupervised Reinforcement Learning Benchmark from Pixels
Sai Rajeswar, Pietro Mazzaglia, Tim Verbelen, Alexandre Piché, Bart Dhoedt, Aaron Courville, Alexandre Lacoste
Reinforcement Learning Tetromino Pixel Reinforcement Learning Benchmark Unsupervised Reinforcement Learning

August 25, 2022

Light-weight probing of unsupervised representations for Reinforcement Learning
Wancong Zhang, Anthony GX-Chen, Vlad Sobal, Yann LeCun, Nicolas Carion
Reinforcement Learning Unsupervised Setting Lightweight High Visual Representation Learning Unsupervised Reinforcement Learning Unsupervised Version Unsupervised Representation

August 4, 2022

Impact Makes a Sound and Sound Makes an Impact: Sound Guides Representations and Explorations
Xufeng Zhao, Cornelius Weber, Muhammad Burhan Hafez, Stefan Wermter
Reinforcement Learning Global Impact Environment Exploration Audio Representation Efficient Exploration Sound Design Sensory Input Unsupervised Reinforcement Learning

May 26, 2022

Unsupervised Reinforcement Adaptation for Class-Imbalanced Text Classification
Yuexin Wu, Xiaolei Huang
Unsupervised Domain Adaptation Domain Invariant Representation Unsupervised Reinforcement Learning

May 23, 2022

POLTER: Policy Trajectory Ensemble Regularization for Unsupervised Reinforcement Learning
Frederik Schubert, Carolin Benjamins, Sebastian Döhler, Bodo Rosenhahn, Marius Lindauer
Model Free Reinforcement Learning Benchmark Unsupervised Reinforcement Learning Policy Regularization Phantom Threat

April 29, 2022

Unsupervised Reinforcement Learning for Transferable Manipulation Skill Discovery
Daesol Cho, Jigang Kim, H. Jin Kim
Robot Manipulation Manipulation Skill Unsupervised Reinforcement Learning Directed Exploration

December 16, 2021

Unsupervised Reinforcement Learning in Multiple Environments
Mirco Mutti, Mattia Mancassola, Marcello Restelli
Reinforcement Learning Unsupervised Reinforcement Learning Policy Gradient Algorithm Exploration Problem