Adaptation in natural and artificial systems
Adaptation in natural and artificial systems
Handwritten Digit Recognition with Binary Optical Perceptron
ICANN '97 Proceedings of the 7th International Conference on Artificial Neural Networks
Training a network with ternary weights using the CHIR algorithm
IEEE Transactions on Neural Networks
| Hi-index | 0.01 |
We present an investigation of the generalization ability of finite size perceptrons with binary couplings. The results for the expected generalization error provide a guide for practical applications by establishing limits for the learning capacity of finite systems. The method applied to find solutions was the genetic algorithm, which showed to be efficient, even for values of @a larger then the Gardner-Derrida storage capacity @a"G"D=1.245, for which the number of solutions is largely reduced. We show that the generalization error of finite size networks for @a up to @a"G"D coincides with the value calculated through the statistical mechanical analysis in the thermodynamic limit.