APPLICATION OF NEURAL COMPUTATION TO CLASSIFICATION OF FOOD HABITS OF MARINES INVERTEBRATES

摘要

In this paper, we apply neural computation to trace ingested food sources of marine invertebrates on inter-tidal flats in Okinawa Island by analyzing their fatty acid profiles. The artificial neural network consists of three layers of neurons, and uses the well-known back-propagation algorithm as the learning method. The neural network classified the food sources with an average 80% success rate.