https://libeldoc.bsuir.by/handle/123456789/54314| Title: | Compressing a convolution neural network based on quantization |
| Authors: | Pertsau, D. Lukashevich, M. Kupryianava, D. |
| Keywords: | материалы конференций;convolution neural network;quantization;quantization-aware training |
| Issue Date: | 2023 |
| Publisher: | BSU |
| Citation: | Pertsau, D. Compressing a convolution neural network based on quantization / D. Pertsau, M. Lukashevich, D. Kupryianava // Pattern Recognition and Information Processing (PRIP'2023) = Распознавание образов и обработка информации (2023) : Proceedings of the 16th International Conference, October 17–19, 2023, Minsk, Belarus / United Institute of Informatics Problems of the National Academy of Sciences of Belarus. – Minsk, 2023. – P. 269–272. |
| Abstract: | Modern deep neural network models contain a large number of parameters and have a significant size. In this paper we experimentally investigate approaches to compression of convolutional neural network. The results showing the efficiency of quantization of the model while maintaining high accuracy are obtained. |
| URI: | https://libeldoc.bsuir.by/handle/123456789/54314 |
| Appears in Collections: | Pattern Recognition and Information Processing (PRIP'2023) = Распознавание образов и обработка информации (2023) |
| File | Description | Size | Format | |
|---|---|---|---|---|
| Pertsau_Compressing.pdf | 431.94 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.