Multiple binary classifiers fusion using induced intuitionistic fuzzy ordered weighted average operator

Combining outputs of a pool of individual classifiers appropriately, as a hot research topic of pattern classification, can generate statistically significant increase in classification accuracy. During the last decades, several fusion algorithms were presented, but few of those focus on two-class classification which possesses widely application area such as sentiment classification, cancer differentiation and so on. Thus this paper presents a multiple binary classifiers fusion scheme which is achieved by the induced intuitionistic fuzzy ordered weighted average (I-IFOWA) operator. Outputs of base classifiers were interpreted as a set of intuitionistic fuzzy values and the fusion procedure is considered as aggregation of the fuzzy information. With different manifestations of the weighting vector, we develop nine specific I-IFOWA operators to implement distinct fusion algorithms, some of which are existing schemes. Experimental results on UCI datasets show that the specific fusion algorithms are effective. Some interesting conclusions are also discussed.