Piece value prediction on Dataset TF GM Classical 2023 (train)
52.87MAE (cp)MLP+CNN
Evaluation results
| Method | Links | |
|---|---|---|
| MLP+CNNCNN encoder layers=4, MLP predictor layers=52026.04 | 52.87 | |
| MLP #2additional hidden layer=true2026.04 | 78.26 | |
| MLP #3non-linear piece location features=rank^2 and file^22026.04 | 78.29 | |
| MLP #1 (Gupta)2026.04 | 78.51 |