Multi Task Reinforcement Learning

Multi-task reinforcement learning (MTRL) aims to train a single agent capable of mastering multiple tasks simultaneously, improving sample efficiency and generalization compared to training separate agents for each task. Current research focuses on addressing challenges like catastrophic forgetting, negative task interference, and sample complexity, often employing modular architectures, federated learning approaches, and algorithms that leverage shared representations or prioritized experience replay to improve performance and stability across diverse tasks. These advancements hold significant promise for enhancing the capabilities of robots and other intelligent agents in real-world applications requiring adaptability and efficient learning from limited data.

Papers

October 4, 2024

Training on more Reachable Tasks for Generalisation in Reinforcement Learning
Max Weltevrede, Caroline Horsch, Matthijs T.J. Spaan, Wendelin Böhmer
Reinforcement Learning Strong Generalization Training Data Policy Reinforcement Learning Multi Task Reinforcement Learning Goal Reachability Go Explore

September 9, 2024

Towards Fast Rates for Federated and Multi-Task Reinforcement Learning
Feng Zhu, Robert W. Heath Jr., Aritra Mitra
Markov Decision Process Policy Gradient Optimal Policy Federated Prompt Cooperation Multi Task Reinforcement Learning Fast Rate Policy Gradient Algorithm

August 20, 2024

Knowledge Sharing and Transfer via Centralized Reward Agent for Multi-Task Reinforcement Learning
Haozhe Ma, Zhengding Luo, Thanh Vinh Vo, Kuankuan Sima, Tze-Yun Leong
Reinforcement Learning Formality Transfer Reward Shaping Multi Task Reinforcement Learning Knowledge Sharing Centralized Reinforcement Learning

July 18, 2024

LIMT: Language-Informed Multi-Task Visual World Models
Elie Aljalbout, Nikolaos Sotirakis, Patrick van der Smagt, Maximilian Karl, Nutan Chen
Multi Task Model Multi Task Reinforcement Learning Robot Policy Learning

June 14, 2024

I Know How: Combining Prior Policies to Solve New Tasks
Malio Li, Elia Piccoli, Vincenzo Lomonaco, Davide Bacciu
Dynamic Environment New Task Prior Knowledge Modularity Function Prior Policy Multi Task Reinforcement Learning Simulated Driving

June 2, 2024

Shared-unique Features and Task-aware Prioritized Sampling on Multi-task Reinforcement Learning
Po-Shao Lin, Jia-Fong Yeh, Yi-Ting Chen, Winston H. Hsu
Feature Extractor Multi Task Reinforcement Learning Task Specific Feature Task Aware Task Sampling

May 25, 2024

Finite-Time Analysis for Conflict-Avoidant Multi-Task Reinforcement Learning
Yudan Wang, Peiyao Xiao, Hao Ban, Kaiyi Ji, Shaofeng Zou
Actor Critic Multi Task Reinforcement Learning Finite Time Finite Time Convergence Dynamic Weighting

May 3, 2024

Natural Policy Gradient and Actor Critic Methods for Constrained Multi-Task Reinforcement Learning
Sihan Zeng, Thinh T. Doan, Justin Romberg
Multi Task Optimal Policy Actor Critic Actor Critic Algorithm Multi Task Reinforcement Learning Natural Policy Gradient Primal Dual Algorithm

April 9, 2024

Efficient Multi-Task Reinforcement Learning via Task-Specific Action Correction
Jinyuan Feng, Min Chen, Zhiqiang Pu, Tenghai Qiu, Jianqiang Yi
Multi Objective Sparse Reward Multi Task Reinforcement Learning Action Correction

January 17, 2024

Sharing Knowledge in Multi-Task Deep Reinforcement Learning
Carlo D'Eramo, Davide Tateo, Andrea Bonarini, Marcello Restelli, Jan Peters
Reinforcement Learning Multi Task Knowledge Based Multi Task Reinforcement Learning Single Task

December 22, 2023

Not All Tasks Are Equally Difficult: Multi-Task Deep Reinforcement Learning with Dynamic Depth Routing
Jinmin He, Kai Li, Yifan Zang, Haobo Fu, Qiang Fu, Junliang Xing, Jian Cheng
New Task Dynamic Routing Routing Problem Multi Task Reinforcement Learning Depth Distribution

December 14, 2023

Less is more -- the Dispatcher/ Executor principle for multi-task Reinforcement Learning
Martin Riedmiller, Tim Hertweck, Roland Hafner
Strong Generalization Generalization Property Design Principle Multi Task Reinforcement Learning Near Optimal Executor

November 19, 2023

Multi-Task Reinforcement Learning with Mixture of Orthogonal Experts
Ahmed Hendawy, Jan Peters, Carlo D'Eramo
Representation Learning Mixture Component Meaningful Representation Multi Task Reinforcement Learning Diverse Representation

November 2, 2023

Contrastive Modules with Temporal Attention for Multi-Task Reinforcement Learning
Siming Lan, Rui Zhang, Qi Yi, Jiaming Guo, Shaohui Peng, Yunkai Gao, Fan Wu, Ruizhi Chen, Zidong Du, Xing Hu, Xishan Zhang, Ling Li, Yunji Chen
Multi Task Temporal Attention Reinforcement Learning Benchmark Multi Task Reinforcement Learning Contrastive Learning Module

November 1, 2023

Federated Natural Policy Gradient and Actor Critic Methods for Multi-task Reinforcement Learning
Tong Yang, Shicong Cen, Yuting Wei, Yuxin Chen, Yuejie Chi
Multi Task Policy OpTimization Federated Reinforcement Learning Multi Task Reinforcement Learning Tabular Markov Decision Process

October 23, 2023

Enhancing Robotic Manipulation: Harnessing the Power of Multi-Task Reinforcement Learning and Single Life Reinforcement Learning in Meta-World
Ghadi Nehme, Ishan Sabane, Tejas Y. Deo
Real Power Proximal Policy Optimization Robotic Manipulation Multi Task Reinforcement Learning Soft Actor Critic Single Agent Reinforcement Learning Trust Region Policy Optimization Possible World

October 20, 2023

Provable Benefits of Multi-task RL under Non-Markovian Decision Making Processes
Ruiquan Huang, Yuan Cheng, Jing Yang, Vincent Tan, Yingbin Liang
Markov Decision Process Multi Task Learning Multi Task Near Optimal Policy Multi Task Reinforcement Learning Provable Benefit

October 12, 2023

Cross-Episodic Curriculum for Transformer Agents
Lucy Xiaoyang Shi, Yunfan Jiang, Jake Grigsby, Linxi "Jim" Fan, Yuke Zhu
Cross Attention Mechanism Multi Task Reinforcement Learning Transformer Agent

September 27, 2023

Distill Knowledge in Multi-task Reinforcement Learning with Optimal-Transport Regularization
Bang Giang Le, Viet Cuong Ta
Knowledge Distillation Multi Task Learning Optimal Transport Multi Task Reinforcement Learning

September 15, 2023

Projected Task-Specific Layers for Multi-Task Reinforcement Learning
Josselin Somerville Roberts, Julia Di
Multi Task Reinforcement Learning

Multi Task Reinforcement Learning

Papers

Training on more Reachable Tasks for Generalisation in Reinforcement Learning

Towards Fast Rates for Federated and Multi-Task Reinforcement Learning

Knowledge Sharing and Transfer via Centralized Reward Agent for Multi-Task Reinforcement Learning

LIMT: Language-Informed Multi-Task Visual World Models

I Know How: Combining Prior Policies to Solve New Tasks

Shared-unique Features and Task-aware Prioritized Sampling on Multi-task Reinforcement Learning

Finite-Time Analysis for Conflict-Avoidant Multi-Task Reinforcement Learning

Natural Policy Gradient and Actor Critic Methods for Constrained Multi-Task Reinforcement Learning

Efficient Multi-Task Reinforcement Learning via Task-Specific Action Correction

Sharing Knowledge in Multi-Task Deep Reinforcement Learning

Not All Tasks Are Equally Difficult: Multi-Task Deep Reinforcement Learning with Dynamic Depth Routing

Less is more -- the Dispatcher/ Executor principle for multi-task Reinforcement Learning

Multi-Task Reinforcement Learning with Mixture of Orthogonal Experts

Contrastive Modules with Temporal Attention for Multi-Task Reinforcement Learning

Federated Natural Policy Gradient and Actor Critic Methods for Multi-task Reinforcement Learning

Enhancing Robotic Manipulation: Harnessing the Power of Multi-Task Reinforcement Learning and Single Life Reinforcement Learning in Meta-World

Provable Benefits of Multi-task RL under Non-Markovian Decision Making Processes

Cross-Episodic Curriculum for Transformer Agents

Distill Knowledge in Multi-task Reinforcement Learning with Optimal-Transport Regularization

Projected Task-Specific Layers for Multi-Task Reinforcement Learning