Dual Reinforcement Learning

Dual reinforcement learning (RL) tackles complex problems by decomposing them into coupled sub-problems, each addressed by a separate RL agent or policy. Current research focuses on developing efficient algorithms, such as those based on policy mirror descent or Deep Q-Networks (DQNs), to optimize these dual policies, often leveraging techniques like Bregman divergences for improved convergence. This approach finds applications in diverse fields, from optimizing resource allocation in real-world systems like bike-sharing to enhancing the performance of recommender systems and enabling more robust robot manipulation through imitation learning. The resulting improvements in efficiency and performance demonstrate the significant potential of dual RL for solving challenging sequential decision-making problems.

Papers

October 2, 2024

Dual Approximation Policy Optimization
Zhihan Xiong, Maryam Fazel, Lin Xiao
Bregman Divergence Policy Mirror Descent Policy Representation Dual Reinforcement Learning

June 2, 2024

Dual Policy Reinforcement Learning for Real-time Rebalancing in Bike-sharing Systems
Jiaqi Liang, Defeng Liu, Sanjay Dominik Jena, Andrea Lodi, Thibaut Vidal
Bike Sharing System Dynamic Rebalancing Vehicle Rebalancing Dual Reinforcement Learning

June 7, 2023

Dual policy as self-model for planning
Jaesung Yoo, Fernanda de la Torre, Guangyu Robert Yang
Self Training Model Free Model Free Policy Self Model Dual Reinforcement Learning Hybrid Policy

February 21, 2023

Dual Policy Learning for Aggregation Optimization in Graph Neural Network-based Recommender Systems
Heesoo Jung, Sangpil Kim, Hogun Park
Graph Neural Network Knowledge Graph GNN Model Recommendation Task GNN Based Optimal Aggregation Dual Reinforcement Learning

February 16, 2023

Dual RL: Unification and New Methods for Reinforcement and Imitation Learning
Harshit Sikchi, Qinqing Zheng, Amy Zhang, Scott Niekum
Reinforcement Learning Imitation Learning Offline Reinforcement Learning Positive Reinforcement New Method Anti Unification Dual Reinforcement Learning

March 18, 2022

Goal-conditioned dual-action imitation learning for dexterous dual-arm robot manipulation
Heecheol Kim, Yoshiyuki Ohmura, Yasuo Kuniyoshi
Imitation Learning Dual Arm Robot Dexterous Robotic Manipulation Bimanual Robot Dual Reinforcement Learning Goal Conditioned Imitation

March 4, 2022

Self-Supervised Learning for Joint Pushing and Grasping Policies in Highly Cluttered Environments
Yongliang Wang, Kamal Mokhtar, Cock Heemskerk, Hamidreza Kasaei
Self Supervised Learning Non Humanoid Robot Functional Grasping Cluttered Environment Deep Reinforcement Learning Approach Stable Pushing Grasp Policy Dual Reinforcement Learning

Dual Reinforcement Learning

Papers

Dual Approximation Policy Optimization

Dual Policy Reinforcement Learning for Real-time Rebalancing in Bike-sharing Systems

Dual policy as self-model for planning

Dual Policy Learning for Aggregation Optimization in Graph Neural Network-based Recommender Systems

Dual RL: Unification and New Methods for Reinforcement and Imitation Learning

Goal-conditioned dual-action imitation learning for dexterous dual-arm robot manipulation

Self-Supervised Learning for Joint Pushing and Grasping Policies in Highly Cluttered Environments