The ChessAIThon project (2025-1-ES01-KA220-VET-000354329) is co-funded by the European Union. The views and opinions expressed in this publication are those of the author(s) only and do not necessarily reflect those of the European Union or the Spanish Service for the Internationalisation of Education (SEPIE). Neither the European Union nor the National Agency SEPIE can be held responsible for them.
Table of Contents
Simplified RL Concept
RL is the process by which the AI learns by continuously playing against itself ( self-play ) and receives a reward (win, +1; draw, 0; loss, -1) at the end of the game, adjusting its decisions to maximize its success. Our project is inspired by this RL architecture but is adapted to a VET context, training the Convolutional Neural Network (CNN) with games played by humans, which reduces the need for intensive self-play resources. Our AI model, ChessMarro, is inspired by the RL architecture of AlphaZero, but adapted to train with human games, which makes it viable in an educational context with limited resources.