Sea clutter neural network classifier: feature selection and MLP design

Authors:
Jose Luis Bárcena-Humanes;David Mata-Moya;María Pilar Jarabo-Amores;Nerea del-Rey-Maestre;Jaime Martín-de-Nicolás
Affiliations:
Signal Theory and Communications Department, Superior Polytechnic School, University of Alcala, Alcala de Henares, Madrid, Spain;Signal Theory and Communications Department, Superior Polytechnic School, University of Alcala, Alcala de Henares, Madrid, Spain;Signal Theory and Communications Department, Superior Polytechnic School, University of Alcala, Alcala de Henares, Madrid, Spain;Signal Theory and Communications Department, Superior Polytechnic School, University of Alcala, Alcala de Henares, Madrid, Spain;Signal Theory and Communications Department, Superior Polytechnic School, University of Alcala, Alcala de Henares, Madrid, Spain
Venue:
IWANN'13 Proceedings of the 12th international conference on Artificial Neural Networks: advances in computational intelligence - Volume Part I
Year:
2013

Citing 3
Cited 0

Neural Networks for Pattern Recognition

Neural Networks for Pattern Recognition
Neural network detectors for composite hypothesis tests

IDEAL'06 Proceedings of the 7th international conference on Intelligent Data Engineering and Automated Learning
Training feedforward networks with the Marquardt algorithm

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

The design of radar detectors in sea clutter environments is really a complex task. A neural network based automatic sea clutter classifier has been designed, as part of an adaptive detector capable of exploiting all the capabilities of detectors designed for specific clutter environments. The most extended sea clutter models have been considered (Gaussian, Weibull and K-distributed). Results show that an MLP with 3 inputs (the variance, the entropy of the modulus of the samples and the correlation coefficient), 6 hidden neurons and 4 outputs, is able to provide a performance similar to the K−NN algorithm with K=10 with a significant reduction in computational cost, a very important feature in real time applications.