Title :
The back thermal symmetry identification by pRAM neural networks
Author :
Ding, Y. ; Guan, Y. ; Clarkson, T.G. ; Clark, R.P.
Author_Institution :
King´´s Coll., London, UK
Abstract :
A pRAM neural network was implemented as a classifier in order to identify the symmetry of thermal patterns of the back. This classifier has a clinical value in the diagnosis of some diseases. The reinforcement self-organising algorithm was adopted which chooses the codebook vectors of the input classes in {0, 1}n hyper-space. Five effective features were extracted as condensed representations of the thermal patterns. A three layer 8-input-pRAM neural net classifier was devised and trained with injected gaussian noise. A promising classification result was obtained and compared with that of the linear Fisher algorithm
Keywords :
Gaussian noise; feature extraction; feedforward neural nets; infrared imaging; learning (artificial intelligence); medical image processing; multilayer perceptrons; pattern classification; back thermal symmetry identification; codebook vectors; disease diagnosis; feature extraction; hyperspace; injected gaussian noise; linear Fisher algorithm; neural network training; pRAM neural networks; pattern classifier; reinforcement self-organising algorithm; thermal pattern symmetry; thermal patterns; three layer pRAM neural net classifier;
Conference_Titel :
Artificial Neural Networks, 1995., Fourth International Conference on
Conference_Location :
Cambridge
Print_ISBN :
0-85296-641-5
DOI :
10.1049/cp:19950596
