KRAVETS P.I., SHIMKOVICH V.N.
ABSTRACT
The authors propose a method for optimizing the weighting factors of the neural network by genetic algorithm with their implementation on Field-programmable gate array (FPGA). The examples of the hardware and software implementation of setting the weighting factors of neural network using the genetic algorithm by PLIC means are presented. It is shown that the parallelizing of calculations provides a significant acceleration of setting the weighting factors of the neural
network.
KEYWORDS
neural networks, genetic algorithm, FPGA.
REFERENCES
1.Egupov, N.D. (2001), Metody robastnogo, neyro-nechetkogo i adaptivnogo upravleniya [Methods of robust, neuro-fuzzy and adaptive control],Izd-vo MGTU im. Baumana, N.E., Moscow, Russia.
2. Terekhov, V.A. (2002), Neyrosetevye sistemy upravleniya. Ucheb. posobie dlya vuzov [Neural network control systems. Textbook for universities], Vysshaya shkola, Moscow, Russia.
3. Solovyev, V. (2003), Proektirovanie tsifrovykh sistem na osnove PLIS [Design of digital systems based on FPGA], Radio i svyaz, Moscow, Russia.
4. Gilgurt, S.Ya. (2006), “Analysis of the application of reconfigurable FPGA-based calculators for implementing neural networks”, Modelyuvannya ta informatsiyni tekhnolohiyi. Zbirnyk naukovykh prats IPME NAN Ukrayiny. Vypusk 37 [Modeling and Information Technologies. Collected Works], Kyiv, Vol. 37, pp. 168-174.
5. Fletcher, R. and Reeves, C.M. (1964), “Function minimization by conjugate gradients”, Computer Journal, Vol. 7, pp. 149-157.
6. Golovko, V.A. (2001), Neyronnye seti: obuchenie, organizatsiya i primenenie [Neural networks: training, organization and application], IPRZhR, Moscow, Russia.
7. Gorban, A.N. (1990), Obuchenie neyronnykh setey [Training neural networks], SP «ParaGraph», Moscow, Russia.
8. The practical handbook of genetic algorithms. Applications. (2000), Ed. Chambers, L.D., Vol. I, CRC Press, Florida, USA.
9. Subbotin, S.O., Oliynyk, A.O. and Oliynyk, O.O. (2009), Neiteratyvni, evolyutsiyni ta multyagentni metody syntezu nechitkologichnykh i neyromerezhnykh modeley. Pid zah. red. S.O. Subbotina [Not iterative, evolutionary and multi-agent methods of synthesis are not clear logic and neural network models], ZNTU, Zaporozhye, Ukraine.
10. Knut, D.E. (2000), Iskusstvo programmirovaniya. T. 2. Poluchislennye algoritmy [Art of Computer Programming. T. 2. Seminumerical algorithms], Vilyams, Moscow, Russia.
11. Barash, L. (2005), “Algorithm AKS primality testing and constant search for generators of pseudorandom numbers”,Bezopasnost informatsionnykh tekhnologiy, no. 2, pp. 27-38.
12. Kravets, P.I., Shymkovych, V.M. and Zubenko, G.A. (2012), “Technology hardware and software implementation of artificial neuron and artificial neural networks by means of FPGA”, Visnyk NTUU «KPI». Informatyka, upravlinnya ta obchyslyuvalna tekhnika. Zbirnyk naukovykh prats,no. 55, pp. 174-180.