Chess2vec: Learning Vector Representations for Chess

TL;DR

This paper introduces a novel approach to generate and evaluate vector representations for chess pieces and moves, revealing their latent structure and enabling move prediction from positions.

Contribution

It is the first to generate and evaluate embeddings for chess pieces and moves, and predicts moves from positions, advancing understanding of chess data representations.

Findings

Uncovered latent structure of chess pieces and moves

Predicted chess moves from positions

Preliminary results on neural network embeddings

Abstract

We conduct the first study of its kind to generate and evaluate vector representations for chess pieces. In particular, we uncover the latent structure of chess pieces and moves, as well as predict chess moves from chess positions. We share preliminary results which anticipate our ongoing work on a neural network architecture that learns these embeddings directly from supervised feedback.