Paired Comparisons Method for Solving Multi-Label Learning Problem

摘要: Multi-label classification problem is a further generalization of traditional multi-class learning problem. In multi-label case the classes are not mutually exclusive and any sample may belong to several at same time. Such problems occur in many important applications (in bioinformatics, text categorization, intrusion detection, etc.). this paper we propose new method for solving problem, based on paired comparisons approach. each pair possibly overlapping separated by two probabilistic binary classifiers, which isolate non-overlapping areas. Then individual probabilities generated classifiers combined together estimate final class fitting extended Bradley-Terry model with ties. Experimental performance evaluation well-known benchmark datasets has demonstrated outstanding accuracy results proposed method.