Unlabeled Data Driven Channel-Wise Bit-Width Allocation and Quantization Refinement | |
Yong Yuan1,2; Chen Chen1,2; Xiyuan Hu1,2; Silong Peng1,2,3 | |
2019 | |
会议日期 | 2019.12.12-15 |
会议地点 | Sydney, Australia |
英文摘要 | Network quantization can effectively reduce computation and memory costs, facilitating the deployment of complex Deep Neural Networks (DNNs) on mobile equipment. However, the low-bit quantization without time-consuming training or access to the full datasets is still a challenging problem. In this paper, we develop a two-stage quantization method to address these issues, which only requires a few unlabeled samples. Firstly, we present a gradient-based approach to analyze per-channel sensitivity and optimize the bit-width allocation for different channels according to their sensitivity. Secondly, we propose to refine the quantization model by distilling knowledge from the output and intermediate features of the pre-trained model. Extensive experiments on image classification and object detection demonstrate the effectiveness of the proposed method, and it can achieve a promising result in 4-bit quantization. |
内容类型 | 会议论文 |
源URL | [http://ir.ia.ac.cn/handle/173211/25818] |
专题 | 自动化研究所_智能制造技术与系统研究中心_多维数据分析团队 自动化研究所_个人空间 |
通讯作者 | Chen Chen |
作者单位 | 1.Institute of Automation, Chinese Academy of Sciences, Beijing, China 2.University of Chinese Academy of Sciences, Beijing, China 3.Beijing ViSystem Corporation Limited, China |
推荐引用方式 GB/T 7714 | Yong Yuan,Chen Chen,Xiyuan Hu,et al. Unlabeled Data Driven Channel-Wise Bit-Width Allocation and Quantization Refinement[C]. 见:. Sydney, Australia. 2019.12.12-15. |
个性服务 |
查看访问统计 |
相关权益政策 |
暂无数据 |
收藏/分享 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论