Current Reinforcement Learning

Current reinforcement learning (RL) research focuses on improving the efficiency and robustness of RL agents, particularly in addressing the challenges of limited data, long-term dependencies, and complex, dynamic environments. Active areas include developing offline RL methods that leverage existing datasets, incorporating large language models to improve reward function design and agent behavior, and enhancing model-based RL through improved memory and state representation. These advancements aim to create more adaptable and sample-efficient RL agents with applications ranging from autonomous driving to robotics and beyond.

Papers

May 23, 2024

Offline Reinforcement Learning from Datasets with Structured Non-Stationarity
Johannes Ackermann, Takayuki Osa, Masashi Sugiyama
Data Set High Dimensional Offline Reinforcement Learning Non Stationary Contrastive Predictive Coding Current Reinforcement Learning

May 7, 2024

In-context Learning for Automated Driving Scenarios
Ziqi Zhou, Jingyue Zhang, Jingyuan Zhang, Boyue Wang, Tianyu Shi, Alaa Khamis
Context Learning Automated Driving Reinforcement Learning Agent Semi Supervised Reward Current Reinforcement Learning

March 7, 2024

Mastering Memory Tasks with World Models
Mohammad Reza Samsami, Artem Zholus, Janarthanan Rajendran, Sarath Chandar
State Space Model World Model Long Term Memory Slow Convergence Symbolic Memory Current Reinforcement Learning

September 21, 2023

Curriculum Reinforcement Learning via Morphology-Environment Co-Evolution
Shuang Ao, Tianyi Zhou, Guodong Long, Xuan Song, Jing Jiang
Learning Dynamic Curriculum Reinforcement Learning Co Evolution Morphology Environment Co Evolution SOTA Segmentation Current Reinforcement Learning

January 27, 2023

Outcome-directed Reinforcement Learning by Uncertainty & Temporal Distance-Aware Curriculum Goal Generation
Daesol Cho, Seungjae Lee, H. Jin Kim
Reinforcement Learning High Uncertainty Anticipation Temporal Information Exploration Problem Aware Curriculum Current Reinforcement Learning

Current Reinforcement Learning

Papers

Offline Reinforcement Learning from Datasets with Structured Non-Stationarity

In-context Learning for Automated Driving Scenarios

Mastering Memory Tasks with World Models

Curriculum Reinforcement Learning via Morphology-Environment Co-Evolution

Outcome-directed Reinforcement Learning by Uncertainty & Temporal Distance-Aware Curriculum Goal Generation