Developing and Python Implementation of the Method of Optimal Architecture Prediction Using Neuronal Networks and Graphite Simulator

V. M. Pentkovsky , A. Y. Drozdov , A. D. Golenev , Y. N. Fonin , e-mail: phonin@rambler.ru S. L.Shlykov

A method of prediction of the optimal architecture using neural networks and Graphite simulator is present. A prediction algorithm has been described, also the Python implementation details of algorithms has been described. The results of experiments of searching optimal architectures for several algorithms are given with prediction mistake.

Keywords: architecture, modeling, neural networks, prediction algorithm

pp. 8–15