Technique for Structural Synthesis of Neurons-Converters of Analog- Digital Neural Network

A technique for structural synthesis of neuroelements with complex functions of activation at the coarse-grained level of representation of neural computers is presented. A multistage procedure for synthesis of neurons-converters is given, a functional neuron model as an aggregate of bit operations in the neural-network basis is considered. Basic logical (structural) schemes of neurons-converters are proposed in the form of operational devices. An example of synthesis of the "code-to-time interval" neural-network converter is given using the hardware description language for programmed logical chips. The proposed techniques and technologies can provide the substantial reduction in time period for development of neurons-converters.

References

[1] Loktyukhin V.N., Chelebaev S.V., Antonenko A.V. Unification of a neural network analog-to-digital converter structure invariant to the form of an input signal. Neyrokomp’yutery [Neurocomputers], 2010, no. 4, pp. 54-61 (in Russ.).

[2] Loktyukhin V.N., Chelebaev S.V. A procedure for the synthesis of pulse analog signal converters using neural network operations. Neyrokomp’yutery: Razrab. Primen. [Neurocomputers: Dev. Appl.], 2006, no. 10, pp. 57-71 (in Russ.).

[3] Tank D.W., Hopfield J.J. Simple "neural" optimization networks: an A/D converter, signal decision circuit, and a linear programming circuit. IEEE Trans. Circuits Syst., 1986, vol. CAS-33, no. 5, pp. 533-541.

[4] Lee B.W., Shen B.J. Design of a neural-based A/D converter using modified Hopfield network. IEEE J. Solid-State Circuits, 1989, vol. SC-24, no. 8, pp. 1120-1135.

[5] Avitabile G., Forti M., Manetti S., Marini M. On a class of nonsymmetrical neural networks with application to ADC. IEEE Trans. Circuits Syst., 1991, vol. CAS-38, no. 2, pp. 202-209.

[6] Galushkin A.I. Neyronnye seti: osnovy teorii [Neural networks: fundamentals of the theory]. Moscow, Goryachaya Liniya-Telekom Publ., 2010. 496 p.

[7] Mayorov S.A., Novikov G.I. Printsipy organizatsii tsifrovykh mashin [Principles of organization of digital computers]. Leningrad, Mashinostroenie Publ., 1974. 432 p.

[8] Loktyukhin V.N., Chelebaev S.V., Antonenko A.V. Neyrosetevye analogotsifrovye preobrazovateli [Neural network analog-to-digital converters]. Moscow, Goryachaya Liniya-Telekom Publ., 2010. 128 p.

[9] Smolov V.B., Ugryumov E.P., Artamonov A.B. Vremya-impul’snye vychislitel’nye ustroystva [Time-pulse computing devices]. Moscow, Radio i Svyaz’ Publ., 1983. 288 p.

[10] Loktyukhin V.N. Mikroprotsessory i EVM. Kn. 4. Mikroprotsessornye sistemy sbora i pervichnoy obrabotki impul’sno-analogovoy informatsii [Microprocessors and computers. Book 4. Microprocessor-based systems for the collection and initial processing of pulse-analog information]. Moscow, Energoatomizdat Publ., 2000. 156 p.

[11] Loktyukhin V.N., Chelebaev S.V., Antonenko A.V. Procedures for setting neural network information converters based on programmable VLSI circuits. Vestn. Mosk. Gos. Tekh. Univ. im. N.E. Baumana, Pnborostr. [Herald of the Bauman Moscow State Tech. Univ., Instrum. Eng.], 2009, no. 3, pp. 76-89 (in Russ.).