Abstract
The constant increase in the complexity of artificial neural networks (ANN) is highlighted as one of the trends of their modern development. An analysis of the main approaches to determining the required size of ANN is given. The most interesting approach is based on the reduction of the size of the network. As a criterion that necessitates the exclusion of communication between neurons, the low level of significance of this connection is used. It is shown that the implementation of the principle of changing the structure of ANN “from complex to simple” contradicts the general biological principle of development “from simple to complex”. A biologically based approach to the development of ANN is considered in accordance with the principle “from simple to complex”. As a criterion that necessitates the “birth” of a new neuron, it is proposed to use ambiguity in determining, first of all, the sign of the weighting factor of at least one of its inputs. The need to implement the training procedure is also emphasized on the basis of the above biologically based approach. An example of changing the structure of a neural network (NN) in accordance with the proposed algorithm is considered. The possibility of obtaining ANN with a non-trivial structure, which differs from the frequently used multilayer structure in practice, is underlined.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Russakovsky O, Deng J, Su H, Krause J, Satheesh S, Ma S, Huang Z, Karpathy A, Khosla A, Bernstein M, Berg AC, Fei LF (2015) ImageNet large scale visual recognition challenge. Int J Comput Vision 115(3):211–252
Gu B, Hu F, Liu H (2001) Modelling classification performance for large data sets: an empirical study. Lecture Notes in Computer Science, vol 2118, pp 317–328 (2001)
Han S, Pool J, Tran J, Dally WJ (2015) Learning both weights and connections for efficient neural networks. ArXiv preprint. arXiv:1506.02626
Han S, Mao, H, Dally WJ (2015) Deep compression: Compressing deep neural networks with pruning, trained quantization and Huffman coding. ArXiv preprint. arXiv:1510.00149
Canziani A, Culurciello E, Paszke A (2015) (2017) An analysis of deep neural network models for practical applications. ArXiv preprint. arXiv:1510.00149,arXiv:1605.07678v4
Bhati R, Jain S, Mishra DK, Bhati D (2010) A comparative analysis of different neural networks for face recognition using principal component analysis and efficient variable learning rate. In: IEEE proceeding of 2010 international conference on computer and communication technology (ICCCT), pp 354−359 (2010)
Alyushin MV, Lyubshov AA (2018) The Viola-Jones algorithm performance enhancement for a person’s face recognition task in the long-wave infrared radiation range. In: IEEE proceeding. of the 2018 conference of russian young researchers in electrical and electronic engineering (ElConRus), Moscow and St. Petersburg, Russia, 29 January−1 February 2018, pp 1813−1816
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Kolobashkina, L.V. (2020). Biologically Inspired Algorithm for Increasing the Number of Artificial Neurons. In: Samsonovich, A. (eds) Biologically Inspired Cognitive Architectures 2019. BICA 2019. Advances in Intelligent Systems and Computing, vol 948. Springer, Cham. https://doi.org/10.1007/978-3-030-25719-4_29
Download citation
DOI: https://doi.org/10.1007/978-3-030-25719-4_29
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-25718-7
Online ISBN: 978-3-030-25719-4
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)