Evaluation of artificial neural networks for fungal identification, employing morphometric data from spores of Pestalotiopsis species

Abstract

The relative abilities of the multilayer perceptron, radial basis function, asymmetric radial basis function and learning vector quantization artificial neural networks (ANNs) and two non-neural methods to identify fungal spores were compared. ANNs were trained on morphometric data from spores of Pestalotiopsis spp. and a few species in the related Truncatella and Monochaetia. The optimized neural and statistical classifiers had similar identification success on an unseen data set – between 76-0 and 78-8% of a 16-species group and between 63-0 and 67-7% of a 19-species group. The relative merits of each classifier are discussed, as is the potential of ANNs in mycology.