Optimal Linear Combination for Two-Class Classifiers
Oriol Ramos Terrades, Salvatore Tabbone, Ernest Valveny · International Conference on Advances in Pattern Recognition · 2007
Combining output classifiers has been one of the strategies used to outperform classifications rates on general purpose classification problems. In this paper we propose two linear combination rules that minimize the classifocation error under some constraints in two-class classifiers.