The idea of training Articial Neural Networks to evaluate chess positions has been widely explored in the last ten years. In this paper we investigated dataset impact on chess position evaluation. We created two datasets with over 1.6 million unique chess positions each. In one of those we also included randomly generated positions resulting from consideration of potentially unpredictable chess moves. Each position was evaluated by the Stockfish engine. Afterwards, we created a multi class evaluation model using Multilayer Perceptron. Solution to the evaluation problem was tested with three different data labeling methods and three different board representations. We show that the accuracy for the model trained for the dataset without randomly generated positions is higher than for the model with such positions, for all data representations and 3, 5 and 11 evaluation classes.
Authors
Additional information
- DOI
- Digital Object Identifier link open in new tab 10.1007/978-3-031-30442-2_32
- Category
- Aktywność konferencyjna
- Type
- publikacja w wydawnictwie zbiorowym recenzowanym (także w materiałach konferencyjnych)
- Language
- angielski
- Publication year
- 2023
