Shapovalova S., Moskalenko Y., Baranichenko O.
Increasing the receptive field of neurons of convulsional neural networks

The convolutional neural network architectures for classifying 1D and 2D signals are analyzed. The authors have found that for a high-dimensional input signal, one can ensure an adequate classification accuracy only by using a large number of layers. It is impossible to achieve the required accuracy with limited computing resources. However, if the number of layers is limited, the accuracy decreases, starting from some critical dimensionality value. A method for modifying a convolutional neural network with relatively small number of layers to solve this problem has been proposed. Its effectiveness has been experimentally proved. © 2023, Springer Science+Business Media, LLC, part of Springer Nature.

Keywords: convolutional neural networks, EfficientNet, receptive field, ResNet, WaveNet, Convolutional neural networks, Multilayer neural networks, Network architecture, 1D signals, 2-D signals, Convolutional neural network, Efficientnet, High-dimensional, Neural network architecture, Number of layers, Receptive fields, Resnet, Wavenet, Convolution

Cite:
Shapovalova S., Moskalenko Y., Baranichenko O. (2023). Increasing the receptive field of neurons of convulsional neural networks. Cybernetics and Systems Analysis, 59 (2), 182–189. doi: https://doi.org/10.1007/s10559-023-00568-0 http://jnas.nbuv.gov.ua/article/UJRN-0001392168 [In Ukrainian].