A Comparison of DQN and Dueling DQN in A Super Mario Environment

DOI

10.2991/978-94-6463-370-2_25How to use a DOI?

Keywords

Reinforcement Learning; Deep Q-network; Super Mario Game

Abstract

Reinforcement learning (RL) has made significant advancements in training artificial agents to play video games. Among various RL algorithms, the deep Q-network (DQN) based on Q-learning has shown outstanding performance in this domain. Recently, dueling DQN, as an enhancement to the standard DQN, has garnered significant attention in the research community. However, there remains a need for a comprehensive and detailed comparison between DQN and dueling DQN, specifically in the context of the Super Mario game environment, as well as an examination of their performance differences. This article aims to investigate and compare the advantages and disadvantages of DQN and dueling DQN in the Super Mario game. It seeks to explore the potential reasons underlying the observed differences in their respective performances. The evaluation is conducted over 3000 epochs, during which the final scores achieved by dueling DQN are observed to be slightly higher than those achieved by DQN. By conducting a rigorous and systematic analysis, this research is conducted for improving the understanding of the nuances and performance disparities between DQN and dueling DQN in the specific context of the Super Mario game. The obtained results will shed light on the potential benefits and drawbacks of each algorithm, providing insights for further advancements and improvements in RL-based gaming agents.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Xiangwei Fu
PY  - 2024
DA  - 2024/02/14
TI  - A Comparison of DQN and Dueling DQN in A Super Mario Environment
BT  - Proceedings of the 2023 International Conference on Data Science, Advanced Algorithm and Intelligent Computing (DAI 2023)
PB  - Atlantis Press
SP  - 221
EP  - 231
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6463-370-2_25
DO  - 10.2991/978-94-6463-370-2_25
ID  - Fu2024
ER  -