A NEURAL-NETWORK ARCHITECTURE FOR CLASSIFICATION OF FUZZY INPUTS

A neural network for classification problems with fuzzy inputs is prop osed. A fuzzy input is represented as an LR-type fuzzy set. A generali zed pocket algorithm, called fuzzy pocket algorithm, that uses LR-type fuzzy sets operations and defuzzification method is proposed to train a linear threshold unit (LTU). This LTU node will classify as many fu zzy input instances as possible. Afterward, FV nodes that represent fu zzy vectors will then be generated and expanded, by proposed FVGE lear ning algorithm, to classify those fuzzy input instances that cannot be classified by the LTU node. The similarity degree between FV nodes an d fuzzy inputs is measured by the fuzzy subsethood degree. The network structure is automatically generated. The number of hidden nodes gene rated depends on the overlapping degree of training instances. Besides , on-line learning is supplied, and parameters used are few and insens itive. The relationship between proposed model and hyperbox-based clas sifiers, e.g., Fuzzy ART series and Fuzzy Min-Max series, is also disc ussed. Two sample problems, heart disease and knowledge-based evaluato r, are considered to illustrate the working of the proposed model. The experimental results are very