Action Masking

Action masking in reinforcement learning (RL) involves strategically restricting the set of actions available to an RL agent, improving training efficiency and performance. Current research focuses on applying action masking across diverse domains, including robotics, cybersecurity, and traffic control, often integrating it with techniques like curriculum learning and employing model architectures such as Proximal Policy Optimization (PPO) and Deep Q-Networks (DQNs). This approach addresses challenges posed by large action spaces and safety concerns in real-world applications, leading to more efficient and reliable RL agents for complex tasks. The resulting improvements in data efficiency and performance have significant implications for deploying RL in safety-critical and resource-constrained environments.

Papers

September 13, 2024

Applying Action Masking and Curriculum Learning Techniques to Improve Data Efficiency and Overall Performance in Operational Technology Cyber Security using Reinforcement Learning
Alec Wilson, William Holmes, Ryan Menzies, Kez Smithson Whitehead
Reinforcement Learning Reinforcement Learning Environment RL Agent Data Efficiency Action Masking

June 6, 2024

Excluding the Irrelevant: Focusing Reinforcement Learning through Continuous Action Masking
Roland Stolz, Hanna Krasowski, Jakob Thumm, Michael Eichelbeck, Philipp Gassert, Matthias Althoff
Reinforcement Learning Action Space Action Boundary Action Masking

April 7, 2024

Efficient Reinforcement Learning of Task Planners for Robotic Palletization through Iterative Action Masking Learning
Zheng Wu, Yichuan Li, Wei Zhan, Changliu Liu, Yun-Hui Liu, Masayoshi Tomizuka
Reinforcement Learning Task Planning Robotic System Efficient Reinforcement Learning Action Masking

March 21, 2024

Heuristic Algorithm-based Action Masking Reinforcement Learning (HAAM-RL) with Ensemble Inference Method
Kyuwon Choi, Cheolkyun Rho, Taeyoun Kim, Daewoo Choi
Reinforcement Learning Markov Decision Process Novel Reinforcement Learning Cognitive Heuristic Action Masking

March 29, 2023

Multi-Agent Reinforcement Learning with Action Masking for UAV-enabled Mobile Communications
Danish Rizvi, David Boyle
Multi Agent Reinforcement Learning Unmanned Aerial Vehicle Various Fast Moving Drone Multiple Unmanned Aerial Vehicle Action Masking

July 24, 2022

MAR: Masked Autoencoders for Efficient Action Recognition
Zhiwu Qing, Shiwei Zhang, Ziyuan Huang, Xiang Wang, Yuehuan Wang, Yiliang Lv, Changxin Gao, Nong Sang
Action Recognition Masked Autoencoders Video Recognition Video Model Vanilla Transformer Temporal Redundancy Action Masking

June 21, 2022

Safe and Psychologically Pleasant Traffic Signal Control with Reinforcement Learning using Action Masking
Arthur Müller, Matthia Sabatelli
Reinforcement Learning Action Space Intersection Scenario Traffic Signal Control Traffic Management Action Masking

Action Masking

Papers

Applying Action Masking and Curriculum Learning Techniques to Improve Data Efficiency and Overall Performance in Operational Technology Cyber Security using Reinforcement Learning

Excluding the Irrelevant: Focusing Reinforcement Learning through Continuous Action Masking

Efficient Reinforcement Learning of Task Planners for Robotic Palletization through Iterative Action Masking Learning

Heuristic Algorithm-based Action Masking Reinforcement Learning (HAAM-RL) with Ensemble Inference Method

Multi-Agent Reinforcement Learning with Action Masking for UAV-enabled Mobile Communications

MAR: Masked Autoencoders for Efficient Action Recognition

Safe and Psychologically Pleasant Traffic Signal Control with Reinforcement Learning using Action Masking