| Bain M. and Srinivasan A. (1995). Inductive Logic Programming With Large-Scale Unstructured Data. In D. Michie K. Furukawa and S. Muggleton, editors, Machine Intelligence 14. Oxford University Press.
Muggleton S.H., Bain M.E., Hayes-Michie J. and Michie D. (1989). An experimental comparison of human and machine learning formalisms. Proceedings of the Sixth International Workshop on Machine Learning.
|
| The data provided here concern the chess endgame KRK with three pieces left on the chess board: White King, White Rook and Black King. The problem of learning rules for recognising illegal positions when it is white's turn to move (WTM) was first proposed by [Muggleton S.H., Bain M.E., Hayes-Michie J. and Michie D. (1989)]. This has since become a widely accepted test-bed for ILP systems. The following information is used to learn the concept of illegal position. The examples are represented by the predicate illegal/6 specifying the column and row coorinates of White King, White Rook and Black King, respectively, - the data on ordering and adjacency of rows and columns are provided as a background knowledge using the predicates lt/2 (less_than) and adj/2. Moreover, the predicate krk/7 offers data on the optimal depth of win (number of neccessery moves) in some cases. The first 6 arguments of krk have the same structure as those of illegal/6. The task in this case is to predict optimal length of win. Preliminary results for "win in 0 moves" up to "win in 5 moves" can be found in [Bain 95].
|