Aimed at neuroscientists, coputer scientists, electrical engineers and mathematicians, this book presents the proceedings of the 5th Italian Workshop on Neural Nets. Papers cover subjects such as equivalent architectures of multi-layer perceptrons; evolving ability to respond to time in inhibition in attractor neural networks; holographic networks in econometry; and speaker recognition over telephone lines.
Universal aspects on learning curves and generalization errors, S. Amari; analog VLSI neural networks, M. Valle et al; equivalent architectures of multi-layer perceptrons - complexity issues and approximations, R. Adami et al; evolving ability to respond to time in populations of recurrent networks - a simple model, O. Miglino et al; computational and physical role of local inhibition in attractor neural networks - a simple model, R. Zecchina; backpropagation vector quantization neural network for probability density functions modelling, P. Burrascano; a McCulloch and Pitts network as a learning sensorimotor control system, F.B. Verona and F.E. Lauria; a high performance music composer based on a modified Jordan network, G.P. Drago et al; a neural approach for speaker recognition over telephone lines, A. Paoloni et al; holographic networks in econometry, A.M. Franich et al. (Part contents)