Multi-Pass Q-Networks for Deep Reinforcement Learning with Parameterised Action Spaces

Craig J. Bester, Steven D. James, George D. Konidaris
5/10/2019
cs.LGstat.ML

Official code: cycraig/MP-DQN (240 stars, MIT, last commit 2019-05-17).

Abstract

Parameterised actions in reinforcement learning are composed of discrete actions with continuous action-parameters. This provides a framework for solving complex domains that require combining high-level actions with flexible control. The recent P-DQN algorithm extends deep Q-networks to learn over such action spaces. However, it treats all action-parameters as a single joint input to the Q-network, invalidating its theoretical foundations. We analyse the issues with this approach and propose a novel method, multi-pass deep Q-networks, or MP-DQN, to address them. We empirically demonstrate that MP-DQN significantly outperforms P-DQN and other previous algorithms in terms of data efficiency and converged policy performance on the Platform, Robot Soccer Goal, and Half Field Offense domains.

View on arXivView PDF

Code Implementations(1)

cycraig/MP-DQNOfficial100%
MIT

Source code for the dissertation: "Multi-Pass Deep Q-Networks for Reinforcement Learning with Parameterised Action Spaces"

24054PythonMar 20, 20197 years agoMIT
deep-reinforcement-learningparameterised-action-spacesparameterised-actions

Similar Papers

Action Branching Architectures for Deep Reinforcement Learning

87% match1 repo1222017

Deep Reinforcement Learning in Parameterized Action Space

85% match1 repo792015

Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments

84% match1 repo2.8k2017

Soft Actor-Critic for Discrete Action Settings

84% match1 repo5.9k2019

Soft Decomposed Policy-Critic: Bridging the Gap for Effective Continuous Control with Discrete RL

84% match2023

SAINT: Attention-Based Policies for Discrete Combinatorial Action Spaces

83% match1 repo12025

Cite this paper

@article{bester2019multipass,
  title  = {Multi-Pass Q-Networks for Deep Reinforcement Learning with Parameterised Action Spaces},
  author = {Craig J. Bester and Steven D. James and George D. Konidaris},
  year   = {2019},
  eprint = {1905.04388},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1905.04388}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1905.04388.svg)](https://codeofpaper.com/paper/1905.04388)

Discussion