COMPARISON OF MULTILAYER PERCEPTRON AND PROBABILISTIC NEURAL NETWORKSIN ARTIFICIAL VISION - APPLICATION TO THE DISCRIMINATION OF SEEDS

In classification problems the most commonly used neural network is pr obably the multilayer perceptron network (MLPN). The probabilistic neu ral network (PNN) is a possible alternative to the MLPN. The PNN is ba sed on the Bayesian approach and a non-parametric estimation of the pr obability density functions of the qualitative classes. In this paper the performances of the PNN and the MLPN were compared on an illustrat ive application which consisted of the discrimination of seed species by artificial vision. The colour images of individual kernels of four species (two cultivated and two adventitious ones) were acquired. A se t of 73 features characterizing the seed size, shape and texture was e xtracted. The data collection was divided into a training set of 1600 seeds and a test set of 800 seeds. A stepwise discriminant analysis ma de it possible to select the first four relevant variables among the 7 3 available ones. The MLPN incorrectly classified 44 and 28 seeds of t he training and test sets respectively. Three configurations of the PN N were tested on the same data collection. The most sophisticated vers ion of the PNN gave 17 and 19 misclassifications in the same data sets . The PNN presents an architecture in which all the units are operatin g in parallel and a hardware implementation of this kind of architectu re is therefore possible. All the scaling parameters of the PNN can be determined from the training set. In contrast, there is no algorithm to automatically determine the structure of the MLPN. (C) 1997 by John Wiley & Sons, Ltd.