https://scholars.lib.ntu.edu.tw/handle/123456789/581331
標題: | Merging deep neural networks for mobile devices | 作者: | Chou Y.-M Chan Y.-M Lee J.-H Chiu C.-Y CHU-SONG CHEN |
關鍵字: | Computer vision; Mergers and acquisitions; Network layers; Neural networks; Convolutional neural network; Feed-forward network; General architectures; High-performance hardware; Inference stages; Resource-limited devices; System development; Training overhead; Deep neural networks | 公開日期: | 2018 | 卷: | 2018-June | 起(迄)頁: | 1767-1775 | 來源出版物: | IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops | 摘要: | In this paper, a novel method to merge convolutional neural networks for the inference stage is introduced. When two feed-forward networks already trained for handling different tasks are given, our method can align the layers of these networks and merge them into a unified model by sharing the representative weights. The performance of the merged model can be restored or improved via re-training. Without needing high-performance hardware, the proposed method effectively produces a compact model to run the original tasks simultaneously on resource-limited devices. The system development time, as well as training overhead, is substantially reduced because our method leverages the co-used weights and preserves the general architectures of the well-trained networks. The merged model is jointly compressed and can be implemented faster than the original models with a comparable accuracy. When combining VGG-Avg and ZF-Net models, our approach can achieve higher than 12 and 2.5 times of compression and speedup ratios compared to the original whole models, respectively, while the accuracy remains approximately the same. ? 2018 IEEE. |
URI: | https://www.scopus.com/inward/record.uri?eid=2-s2.0-85060861655&doi=10.1109%2fCVPRW.2018.00220&partnerID=40&md5=0c85f9165976ba3c1bc476a49b1bfa2c https://scholars.lib.ntu.edu.tw/handle/123456789/581331 |
ISSN: | 21607508 | DOI: | 10.1109/CVPRW.2018.00220 |
顯示於: | 資訊工程學系 |
在 IR 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。