I.V. Stepanian – Dr. Sc. (Biol.), Ph.D. (Eng.), Leading Research Scientist, Labor Medicine Institute  of the Russian Academy of Sciences, A.A. Blagonravov Institute for Machine Science of the RAS,  P.I. Chaikovsky Moscow State Сonservatory

E-mail: neurocomp.pro@gmail.com

N.N. Ziep – Post-graduate Student, Department of Electronic Computer, S.A. Lebedev JSC Institute  of Precision Mechanics and Computer Science of RAS (Moscow),  Moscow Institute of Physics and Technology (State University)

E-mail: diepnn83@mail.com

Multilayer convolutional neuron-like structures and algorithms for their dynamic synthesis for pattern recognition problems are investigated. The work relates to the development of the concept of deep learning and is a generalization of the author's technology Agnitus-1, which is based on the elements of the theory of autonomous adaptive control of Zhdanov. Results are presented in the field of training and learning of convolutional bionic neural networks using algorithms for the growth of their structures. Algorithmically synthesized and trained convolutional neural-like networks have stable noise immunity of pattern recognition.

1. LeCun Y., Bengio Y. Convolutional Networks for Images, Speech, and Time-Series, in Ar-bib, M. A. (Eds). The Handbook of Brain Theory and Neural Networks. MIT Press. 1995.
2. ZHdanov A.A. Avtonomnyj iskusstvennyj intellekt. M.: BINOM Laboratoriya znanij. 2008.
3. ZHdanov A.A., Kryzhanovskij M.V., Preobrazhenskij N.B. Bionicheskaya intellektu-al'naya adaptivnaya sistema upravleniya mobil'nym robotom // Iskusstvennyj intellekt. 2002. T. 4. S. 341–350.
4. Hajkin S. Nejronnye seti: polnyj kurs. Izd. 2-e. M.: Izdatel'skij dom «Vil'yams». 2008.
5. Lebedev A.E., ZHdanov A.A. Dinamicheskaya segmentaciya prostranstva priznakov dlya si-stem avtonomnogo adaptivnogo upravleniya i sistem obucheniya s podkrepleniem: Sb. nauchnyh trudov. V 2-h chastyah. CH.1. M.: NIYAU MIFI. 2010. S. 182–190.
6. Stepanyan I.V., Homich A.V., Karpishuk A.V. Princip blochnosti v ehvolyucionnoj opti-mizacii struktur nejronnyh setej // Nejrokomp'yutery: razrabotka i primenenie. 2006. № 3. S. 17–25.
7. Denisov EH.I., Eremin A.L., Stepanyan I.V., Bodyakin V.I. Voprosy izmereniya i ocenki informacionnyh nagruzok pri umstvennom trude // Nejrokomp'yutery: razrabotka, prime-nenie. 2013. № 10. C. 54–63.
8. Schmidhuber J. Deep learning in neural networks: An overview // Neural networks. 2015. V. 61. P. 85–117.
9. Krizhevsky A., Sutskever I., Hinton G.E. Imagenet classification with deep convolutional neural networks // Annual Conference on Neural Information Processing Systems (NIPS). 2012. P. 1106–1114.
10. LeCun Y., Cortes C., Christopher J.C. Tye B. MNIST database of handwritten digits – baza dannnyh rukopisnyh cifr. http://yann.lecun.com/exdb/mnist/