A trial of artificial neural networks for automatically estimating the ageof fish

Artificial neural networks (ANP Ts) have the potential to automate routine ageing of fish with the benefit of increased speed in processing, greater o bjectivity and repeatability of estimates, and a mechanism for quantifying uncertainty of age estimates. ANN models were tested as a means of objectiv ely replicating the age estimates of an experienced human reader. Feed-forw ard back-propagation ANNs, with three layers of neurons (input, hidden and output), were trained to classify the age of previously aged samples of thr ee temperate species. Three ANN structures, where the number of neurons in the hidden layer was varied, were tested for each species. Inputs to each A NN were pixel brightness values along transects across images of sectioned otoliths. The ANN predicted age-class membership by the position of the neu ron in the output layer with the highest value. After training, at least on e of the three ANN structures correctly classified the age of fish from uns een transects for two members of the Sparidae family Acanthopagrus butcheri and Pagrus auratus at an accuracy level approaching that of an expert read er. For a member of the Merlucciidae family, Macruronus novaezelandiae, how ever, which is a species with more complex otolith structure, error rates w ere high for all three ANN structures tested.