Title :
Neural network design using Voronoi diagrams: preliminaries
Author :
Bose, N.K. ; Garga, A.K.
Author_Institution :
Dept. of Electr. & Comput. Eng., Pennsylvania State Univ., University Park, PA, USA
Abstract :
A novel approach based on the construction of a Voronoi diagram is proposed to determine the number of layers, the number of neurons in each layer, and their connection weights for a particular implementation of a neural network. The neural network has a multilayer feedforward topology, and is designed to classify patterns in the multidimensional feature space. To illustrate the procedure, an example is given of the classification of patterns that are not linearly separable in feature space
Keywords :
computational geometry; feedforward neural nets; pattern recognition; Voronoi diagram; multidimensional feature space; multilayer feedforward topology; neural network design; pattern classification; Artificial neural networks; Computational geometry; Feedforward neural networks; Multi-layer neural network; Multidimensional systems; Network topology; Neural networks; Neurons; Nonlinear equations; Systems engineering and theory;
Conference_Titel :
Neural Networks, 1992. IJCNN., International Joint Conference on
Conference_Location :
Baltimore, MD
Print_ISBN :
0-7803-0559-0
DOI :
10.1109/IJCNN.1992.227181
