douzR
Authors: Arash Haratian
Abstract: This game uses a Reinforcement learning algorithm called temporal difference learning(TD).
users play tic-tac-toe with the TD agent and the agent updates its state values.
when the game ends, the precision text above the board will be updated.
Full Description: This game uses a Reinforcement learning algorithm called temporal difference learning (TD learning).
users play tic-tac-toe with the TD agent and the agent updates its state values which are saved in the variable named value_table.
when the game ends, the precision text above the game will be updated, and also performance plot will be updated after some number of games.
all users compete against the same agent (with the same value_table), so the agent will estimate state values more accurately and updates value_table more properly.
the original code for this game is from the first chapter of "Reinforcement Learning: An Introduction, 2nd edition" textbook. you may find more details about the agent and the game (and also ttt.R code) in that chapter.
Keywords: reinforcement learning, temporal difference learning, game, ggplot2, plotly, tic-tac-toe
Shiny app: https://arashharatian.shinyapps.io/douzr
Repo: GitHub - arashHaratian/douzR: online tic-tac-toe using TD method
RStudio Cloud: Posit Cloud
Thumbnail:
Full image:
