Data Clustering with Neural Network Based on Potential Function Generators

Project Objectives

We propose to develop a new strategy of shape-adaptive radial basis functions based on potential functions and optimization procedure for positioning of the cluster centers during the learning process as well as a multilayer neural network architecture with symmetric potential functions and criteria for merging and splitting the clusters. We will analyze how the shape of the clusters changes with dimension of the input data, choice of symmetric potential function and the number of centers.

We will investigate the dependence of our method on these parameters and apply it to several artificial and benchmark data sets in order to study the power of the potential function entities (PFE’s) in generating clustering solutions for various shapes of teaching patterns that are robust with respect to noise in the data.

Several benchmark data sets will be considered to show the clustering performance on the training and test sets achieved by the proposed approach and some other neural network models. The results will be compared with some standard methods for clustering.