https://scholars.lib.ntu.edu.tw/handle/123456789/633719
標題: | Learnable Mixed-precision and Dimension Reduction Co-design for Low-storage Activation | 作者: | Tai, Yu Shan Chang, Cheng Yang Teng, Chieh Fang AN-YEU(ANDY) WU |
關鍵字: | Activation compression | convolutional neural network | dimension reduction | mixed-precision | 公開日期: | 1-一月-2022 | 卷: | 2022-November | 來源出版物: | IEEE Workshop on Signal Processing Systems, SiPS: Design and Implementation | 摘要: | Recently, deep convolutional neural networks (CNNs) have achieved many eye-catching results. However, deploying CNNs on resource-constrained edge devices is constrained by limited memory bandwidth for transmitting large intermediated data during inference, i.e., activation. Existing research utilizes mixed-precision and dimension reduction to reduce computational complexity but pays less attention to its application for activation compression. To further exploit the redundancy in activation, we propose a learnable mixed-precision and dimension reduction co-design system, which separates channels into groups and allocates specific compression policies according to their importance. In addition, the proposed dynamic searching technique enlarges search space and finds out the optimal bit-width allocation automatically. Our experimental results show that the proposed methods improve 3.54%/1.27% in accuracy and save 0.18/2.02 bits per value over existing mixed-precision methods on ResNet18 and MobileNetv2, respectively. |
URI: | https://scholars.lib.ntu.edu.tw/handle/123456789/633719 | ISBN: | 9781665485241 | ISSN: | 15206130 | DOI: | 10.1109/SiPS55645.2022.9919207 |
顯示於: | 電機工程學系 |
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。