Distinctive Feature Extraction for Fast and Reliable Classification in Complex Systems

In this work, an approach for establishment of class membership in complex systems is reported. The classification is based on adaptive recognition facilitating the discovery of pattern features that make them distinct from objects belonging to different classes. By viewing a pattern as a representation of extracts of information regarding various features of an object, most traditional recognition methods tend to achieve categorization by identifying the resemblances amongst the class members. In this work, a different view of classification is presented. The classification is based on identification of distinctive features of patterns. It argues that the basic functioning of the established methods also implies that the members of different classes have different values for some or all of such features expressing the objects under consideration. That is, the categorization can also be based on recognition of dissimilarities and distinctions between the objects fitting in different classes. Our proposed approach in based on identifying such charactering dissimilarities, which will then form the distinctive features of patterns and objects. In other words, objects are classified as members of a particular class if they possess some features, which make them distinguished from other objects present in the universe of objects. The proposed approach and its language work in a general manner. Consequently, the corresponding codes can be developed and utilized as a general adaptive pattern recognition scheme. The generality of the approach proposed in this work, makes it applicable to many classification and pattern recognition problems encountered in complex systems.