Adaptive classifiers using ontogenetic neural networks with feedback

The authors give the results of software simulation of Hebbian (D. O. Hebb, 1949) associative learning. Ontogenesis in feedback neural networks similar to those devised by Hopfield implies an initial structure and a plan for associative learning with growth. Experimental results are given to show that the advantages of ontogenetic neural networks configured as adaptive classifiers include rapid adaptation, good performance in classifying correctly, and the ability to cope with high levels of noise in training and operation.<>