Neural networks in biological taxonomy

The need for constant improvements in the classification of living organisms brings to biology an increased need for mathematical and computational tools. This need is being addressed at the Departments of Computer Sciences and Biochemistry, Universidade Federal de Minas Gerais, Brazil, by the development of neural network programs directed towards the identification of living organisms. Neural networks are very efficient at solving identification problems because they are highly flexible and insensitive to the effects of missing data and noise. This paper presents Neurotaxon, a general taxonomy tool based on back-propagation networks, and describes some procedures for achieving maximum accuracy in the taxonomical classification of living organisms.