AN EVOLVING CASCADE SYSTEM BASED ON NEURO-FUZZY NODES

An evolving cascade system based on fuzzy-neurons and its learning procedures are proposed in the paper. During a learning procedure in an online mode, the proposed system tunes both its parameters and its architecture. Neuro-fuzzy systems are proposed as nodes of the evolving cascade system. A method based on the gradient procedure of a learning criterion minimization is proposed for membership functions’ tuning in the neuro-fuzzy nodes. Synaptic weights, centers and width parameters of the membership functions are tuned during the learning procedure. Software that implements the proposed evolving cascade neuro-fuzzy system’s architecture has been developed. A number of experiments has been held in order to research the proposed system’s properties. Experimental results have proven the fact that the proposed system could be used to solve a wide range of Data Mining tasks. Data sets are processed in an online mode. The proposed system provides computational simplicity, and data sets are processed faster due to the possibility of parallel tuning for the evolving cascade system. A distinguishing feature of the proposed system is that there is no need of a large training set for the system to be tuned.