Active Reinforcement Learning

Active reinforcement learning (ARL) aims to improve the efficiency and robustness of reinforcement learning by strategically selecting data for training. Current research focuses on developing algorithms that actively explore environments, prioritizing data acquisition to enhance generalization and reduce the need for extensive, potentially costly, trial-and-error learning. This involves techniques like uncertainty-aware neural networks and novel reward structures that guide exploration towards informative states, improving performance in specific target environments while maintaining robustness to variations. ARL's impact spans diverse fields, from optimizing building energy consumption to personalized healthcare monitoring, by enabling more efficient and effective learning in complex, real-world scenarios.

Papers

December 16, 2023

Active Reinforcement Learning for Robust Building Control
Doseok Jang, Larry Yan, Lucas Spangher, Costas Spanos
Reinforcement Learning Reinforcement Learning Agent Robot Control Unsupervised Environment Design Active Reinforcement Learning

September 18, 2023

Actively Learning Reinforcement Learning: A Stochastic Optimal Control Approach
Mohammad S. Ramadan, Mahmoud A. Hayajnh, Michael T. Tolley, Kyriakos G. Vamvoudakis
Reinforcement Learning Active Exploration Stochastic Optimal Control Stochastic Control Active Reinforcement Learning

June 1, 2023

Active Vision Reinforcement Learning under Limited Visual Observability
Jinghuan Shang, Michael S. Ryoo
Full Observability Active Vision Active Reinforcement Learning Reward Consistency

April 28, 2023

Active Reinforcement Learning for Personalized Stress Monitoring in Everyday Settings
Ali Tazarv, Sina Labbaf, Amir Rahmani, Nikil Dutt, Marco Levorato
Fine Grained Physiological Signal Wearable Sensor Daily Living Stress Biomarkers Active Reinforcement Learning

December 19, 2022

Near-optimal Policy Identification in Active Reinforcement Learning
Xiang Li, Viraj Mehta, Johannes Kirschner, Ian Char, Willie Neiswanger, Jeff Schneider, Andreas Krause, Ilija Bogunovic
Active Exploration Near Optimal Policy Real World Reinforcement Learning Contextual Bayesian Optimization Policy Identification Active Reinforcement Learning

January 11, 2022

Active Reinforcement Learning -- A Roadmap Towards Curious Classifier Systems for Self-Adaptation
Simon Reichhuber, Sven Tomforde
Reinforcement Learning Active Learning Classification Model Reinforcement Learning Approach Requirement Based Roadmap Self Adaptation Reactive Behavior Active Reinforcement Learning