A method of solving pattern or image recognition problems by learning Boolean formulas
Ventzeslav Valev, Petia I. Radeva · 2003
A method of solving supervised pattern recognition problems based on the model of learning Boolean formulas is suggested. It is proved that this method of learning is of NP-complexity. An efficient learning procedure using some tools of combinatorics and graph theory is proposed. The suggested method differs from those known in its diminished number of computational operations. The results obtained are applied to supervised image recognition problems.>