Relation between VGA-classifier and MLP : determination of network architecture

• Autorzy: Bandyopadhyay S. , Pal S.K.
• Języki publikacji: EN

Abstrakty

EN

An analogy between a genetic algorithm based pattern classification scheme (where hyper-planes are used to approximate the class boundaries through searching) and multiplayer per-ceptron (MLP) based classifier is established. Based on this, a method for determining the MLP architecture automatically is described. It is shown that the architecture would need at-most two hidden layers, the neurons of which are responsible for generating hyperplanes and regions. The neurons in the second hidden and output layers perform the AND & OR func-tions respectively. The methodology also includes a post processing step which automatically removes any redundant neuron in the hidden/output layer. An extensive comparative study of the performance of the MLP, thus derived using the proposed method, with those of several other conventional MLP's is presented for different data sets.

Słowa kluczowe

EN

hyperplane fitting; boundary approximation; hard limiting neuron; network architecture design; variable string length genetic algorithm

• Wydawca: IOS Press
• Czasopismo: Fundamenta Informaticae
• Rocznik: 1999
• Tom: Vol. 37, Nr 1,2
• Strony: 177--199
• Opis fizyczny: tab., wykr., bibliogr. 19 poz.

Twórcy

autor

Bandyopadhyay S.

autor

Pal S.K.

• Machine Intelligence Unit, Indian Statistical Institute 203 B.T. Road, Calcutta 700 035, India,