This article defines the challenge of training neural-networks on specific chess endgames and benchmarking their efficacy against the existing sub-8-man endgame tables. The key tasks are to measure how well they play and to infer higher-order rules and guidelines for play from them.
Paper
Full text
Chess endgame news: an endgame challenge for neural nets
Semantic Scholar · Computer Science · 2019
Abstract
This article defines the challenge of training neural-networks on specific chess endgames and benchmarking their efficacy against the existing sub-8-man endgame tables. The key tasks are to measure how well they play and to infer higher-order rules and guidelines for play from them.