Akademik Veri Yönetim Sistemi
Journal of Advanced Computational Intelligence and Intelligent Informatics, cilt.5, sa.1, ss.37-43, 2001 (Scopus)
This study presents a theoretical investigation of the rank-basrd multiple classifier decision problem for closed-set pattern identification. The problem of combining the decisions of more than one classifiers with raw outputs in the form of candidate class rankings is considerd and formulated as a general discrete optimization problem with an objective function based on the total probability of correct Decision. This formulation uses certain performance statistics about the joint behaviour of the ensemble of ciassifiers, which need to be estimated from cross-validation Data. An initial approch leads to an integer (binary) programming problem with a simple and global optimum so lution but of prohibitive dimensionality. Therefore, we present a partitioning formalism under whihn this dimensionality can be reduced by incorporating our prior knowledge about the prblem domain and the structure of the traning data. it is also shown that formalism can effectively expiam a number ot successfully used combination approaches in the literature.