Journal of Science and Science Education

  • P-ISSN3022-8719

Article Detail

Home > Article Detail
  • P-ISSN 3022-8719

Prev

Next

Vol.49, No.2
PDF Citation Share

Concept of Artificial Intelligence Reinforcement Learning and Applications to Experimental Data Analysis

Journal of Science and Science Education / Journal of Science and Science Education, (P)3022-8719;
2024, v.49 no.2, pp.66-74
https://doi.org/10.23335/JSSE.2024.49.2.66


& (2024). Concept of Artificial Intelligence Reinforcement Learning and Applications to Experimental Data Analysis. , 49(2), 66-74, https://doi.org/10.23335/JSSE.2024.49.2.66
  • Downloaded
  • Viewed

Abstract

Reinforcement learning (RL) is a method that addresses sequential decision-making problems by enabling an agent to interact with an environment and learn from rewards. Deep RL, a fusion of artificial neural networks and RL, shows promise in surpassing the constraints of supervised and unsupervised learning in machine learning. This study delves into the policy iteration learning process of RL using dynamic programming. It elaborates on how the value function and Q-function, derived from the Bellman equation, are leveraged in a Grid World environment to elucidate the core tenets of RL. Furthermore, practical applications of deep RL are showcased through the utilization of the A3C (Asynchronous Advantage Actor-Critic) algorithm in the analysis of Extended X-ray Absorption Fine Structure (EXAFS). This demonstration underscores the effective integration of deep RL in scientific data analysis.

keywords
Reinforcement learning, Machine learning, Deep learning, Artificial intelligence, EXAFS


  • Downloaded
  • Viewed
  • 0KCI Citations
  • 0WOS Citations

Other articles from this issue

Recommanded Articles

상단으로 이동

Browse Articles

Info

Editorial Policy

Bulletin Board

  This website recommended the use   the Chrome browser for journal website.

Journal of Science and Science Education

e-Submission OA Policy