Article

Article title SYNTHESIS CLASS LIBRARY OBJECTIVE-C_NEURON FOR THE ARTIFICIAL NEURAL NETWORKS IN PATTERN RECOGNITION
Authors N.I. Vitiska, S.K. Bukhantseva
Section SECTION III. ARTIFICIAL INTELLECT AND INDISTINCT SYSTEMS
Month, Year 07, 2014 @en
Index UDC 004.421.6
DOI
Abstract Describes the object-oriented method for creating libraries in the implementation of artificial neural networks in mobile and portable devices, offering software-based models of pattern recognition. This article addressed the challenges of modeling of neural networks for various applications complexly using by standard methods. Methodology of designing neural networks for iOS platform was proposed. Also methods of creating and learning through the developed library Objective-C_Neuron were created that give ability to expand the toolkit. Particular attention has been paid to increase range of application of neural networks. The main advantage of proposed system is a comfy way of multithreading working. Such applications work on mobile devices and personal computers. The possibility of parallelization basic calculations to establish the connections between neurons, where the calculation of weights will be implemented in GUP. Task graphs assume independent parallel processing, and GPU originally multipotochen. Microarchitecture is designed to exploit the available large number of threads that require execution. Thus, non-graphical calculations implemented on the GPU give a significant increase in performance compared to traditional solutions.

Download PDF

Keywords Neural network; methodology; method; library; toolkit; multithreading.
References 1. Chernukhin Yu.V. Iskusstvennyy intellekt i neyrokompyutery [Artificial intelligence and neural computers]. Taganrog: Izd-vo TRTU, 1997, 273 p.
2. Kirsanov Eh.Yu. Neyrokompyutery s parallelnoy arkhitekturoy [Neural computers with parallel architecture]. Moscow: Radio-tekhnika, 2004, 496 p.
3. Zlobin V.K., Ruchkin V.N. Neyroseti i neyrokompyutery [Neural networks and neurocomputers]: Ucheb. posobie. Saint-Petersburg: BKhI – Peterburg, 20011, 256 p.
4. Galushkin A.I. Teoriya neyronnykh setey [The theory of neural networks]. Moscow: IPRZh «Radiotekhnika», 2000, 416 p.
5. Extensible Markup Language. (XML) 1.0. W3C Recommendation. Available at: //http://www.w3.org/TR.
6. The SGML/XML Web Page-Extensible Markup Language (XML). Available at:
//http://www.oasis-open.org/cover/xml.html.
7. Simard P.Y. Best Practices for Convolutional Neural Networks Applied to Visual Document Analysis, Microsoft, pp. 23-24.

Comments are closed.