Classification Using Multi-valued Pulse Coupled Neural Network

Authors:
Xiaodong Gu
Affiliations:
Department of Electronic Engineering, Fudan University, Shanghai, China 200433
Venue:
Neural Information Processing
Year:
2008

Citing 9
Cited 0

Image thinning using pulse coupled neural network

Pattern Recognition Letters
Feature linking via synchronization among distributed assemblies: Simulations of results from cat visual cortex

Neural Computation
Global icons and local icons of images based unit-linking PCNN and their application to robot navigation

ISNN'05 Proceedings of the Second international conference on Advances in neural networks - Volume Part II
PCNN models and applications

IEEE Transactions on Neural Networks
Physiologically motivated image fusion for object detection using a pulse coupled neural network

IEEE Transactions on Neural Networks
Perfect image segmentation using pulse coupled neural networks

IEEE Transactions on Neural Networks
Finding the shortest path in the shortest time using PCNN's

IEEE Transactions on Neural Networks
Foveation by a pulse-coupled neural network

IEEE Transactions on Neural Networks
Image shadow removal using pulse coupled neural network

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper introduces how to use multi-valued PCNN (Pulse Coupled Neural Network) proposed in this paper to do classification. 2-dimensional data can be projected onto two-dimensional PCNN locally laterally linked. Different pulse waves generated by training data label different regions corresponding to different classes. The same pulse wave labels the region corresponding to the same class. Meeting of different pulse waves obtains the separatrixes of different classes. In order to differentiate different pulse waves, outputs of neurons in PCNN should be multi-valued. We call networks composed of these neurons multi-valued PCNNs. The number of classes determines the number of output value of each neuron. N-valued PCNN can be used to classify N-1different classes. Experimental results of the 2-dimensional salmon-weever classification show that the correct recognition rate of test set is 98.11% (3477/3544) when training samples are only 10% of all samples.