Automatic clustering of generalized regression neural network by similarity index based fuzzy c-means clustering

Abstract

In general regression neural networks (GRNN), one drawback is that the number of training vectors is proportional to the number of hidden nodes, thus a large number of training vectors produce a larger architecture, which is a major disadvantage for many applications. In this paper we proposed an efficient clustering technique referred to as 'similarity index fuzzy c-means clustering'. This technique uses the conventional fuzzy c-means clustering preceded by a technique based on similarity indexing to automatically cluster input data which are relevant to the system. The technique employs a one-pass similarity measures on the data to calculate the similarity index. This index indicates the degree of similarity in which data is clustered. Similar data then undergoes fuzzy c-means iterative process to determine their cluster centers. We applied the technique for system identification and modeling and found the results to be encouraging and efficient. This algorithm offers better performance than conventional algorithm which using energy only. The vocabulary for the experiment includes English digit from 1 to 9. These experimental results were conducted by 360 utterances from a male speaker. Experimental results show that the accuracy of the algorithm is quite acceptable.