The known classical architectures of neural networks have many weak properties and high limitations such as great difficulties in choosing proper parameters (the numbers of neurons, connections, layers), in learning and in expanding a learned network and some others. In this paper, to overcome these shortcomings, we consider the architecture of a neural network in which the recognition is performed in pairs. The neural network is created on the basis of the neural network architecture that implements the method of the nearest neighbor, without using analytical expressions and a set of selected samples. The paper shows that such an architecture can be applied to the recognition problem with a very large number of images (classes). At the same time, the proposed neural network has a simple architecture, with the possibility of simpler learning of the neural network. The neural network architecture allows to add new recognizable images to the neural network without changing the previous network settings.
Download file
Counter downloads: 289
- Title The architecture of a neural network with a sequential division of images into pairs
- Headline The architecture of a neural network with a sequential division of images into pairs
- Publesher
Tomsk State University
- Issue Prikladnaya Diskretnaya Matematika - Applied Discrete Mathematics 41
- Date:
- DOI 10.17223/20710410/41/10
Keywords
архитектуры нейронных сетей, определяемые нейронные сети, нейрокопьютер, сверточные сети, алгоритмы обучения нейронных сетей, neural network architectures, neural networks, neurocomputer, convolu-tional networks, neural network training algorithmsAuthors
References
Барский А. Б. Нейронные сети: распознавание, управление, принятие решений. М.: Финансы и статистика, 2004.
Головко В. Л. Нейронные сети: обучение, организация и применение / под ред. А. И. Галушкина. Кн. 4. М.: ИПРЖР, 2001.
Уоссермэн Ф. Нейрокомпьютерная техника. Теория и практика. М.: Мир, 1992.
LeCun Y., Bengio Y., and Hinton G. Deep learning // Nature. 2015. No. 521. P. 436-444.
Schmidhuber J. Deep learning in neural networks: An overview // Neural Networks. 2015. No. 61. P. 85-117.
http://www.neuralnetworksanddeeplearning.com/index.html - Neural Networks and Deep Learning.
Geidarov P. Sh. Neural networks on the basis of the sample method // Automatic Control and Computer Sci. 2009. V.43. No. 4. P. 203-210.
Geidarov P. Sh. Multitasking application of neural networks implementing metric methods of recognition // Autom. Remote Control. 2013. V. 74. No. 9. P. 1474-1485.
Биргер И. А. Техническая диагностика. M.: Машиностроение, 1978.
Гейдаров П. Ш. Алгоритм реализации метода ближайшего соседа в многослойном персептроне // Труды СПИИРАН. 2017. Вып. 51. С. 123-151.
Васин Д. Ю., Аратский А. В. Распознавание символов на основе инвариантных моментов графических изображений //25 Междунар. конф. GraphiCon 2015. М., 2015. С. 259-264.

The architecture of a neural network with a sequential division of images into pairs | Prikladnaya Diskretnaya Matematika - Applied Discrete Mathematics. 2018. № 41. DOI: 10.17223/20710410/41/10
Download full-text version
Counter downloads: 574