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
Reference Example
If a teacher needs to show how this complex encoding is done, they can reference the practical example of mapping the move "e2e4" to index 820 (document 'Codificacion_Movimiento_e2e4_820.pdf'). This illustrates how a complex chess instruction is reduced to a single number that the AI can easily predict.
The "Referee" Algorithm (Boolean Mask)
Once the CNN produces a vector of 4096 probabilities , it is likely that some of the highest predictions correspond to illegal moves. To prevent this, the concept of a Boolean Mask of legal moves is used, a strategy employed in the final implementation of ChessMarro. The 77x8x8 matrix input to the network includes 64 layers that defines all possible moves.