Multi-Granularity Pruning for Model Acceleration on Mobile Devices | |
Zhao TL(赵天理)5,6; Zhang X(张希)5,6; Zhu WT(朱文涛)4; Wang JX(王家兴)3; Yang S(杨森)1; Liu J(刘季)2; Cheng J(程健)5,6 | |
2022 | |
会议日期 | 2022-07 |
会议地点 | 线上 |
关键词 | Deep Neural Networks Network Pruning Structured Pruning Non-structured Pruning Single Instruction Multiple Data |
英文摘要 | For practical deep neural network design on mobile devices, it is essential to consider the constraints incurred by the computational resources and the inference latency in various applications. Among deep network acceleration approaches, pruning is a widely adopted practice to balance the computational resource consumption and the accuracy, where unimportant connections can be removed either channel-wisely or randomly with a minimal impact on model accuracy. The coarse-grained channel pruning instantly results in a significant latency reduction, while the fine-grained weight pruning is more flexible to retain accuracy. In this paper, we present a unified framework for the Joint Channel pruning and Weight pruning, named JCW, which achieves a better pruning proportion between channel and weight pruning. To fully optimize the trade-off between latency and accuracy, we further develop a tailored multi-objective evolutionary algorithm in the JCW framework, which enables one single round search to obtain the accurate candidate architectures for various deployment requirements. Extensive experiments demonstrate that the JCW achieves a better trade-off between the latency and accuracy against previous state-of-the-art pruning methods on the ImageNet classification dataset. |
语种 | 英语 |
内容类型 | 会议论文 |
源URL | [http://ir.ia.ac.cn/handle/173211/52090] |
专题 | 类脑芯片与系统研究 |
通讯作者 | Cheng J(程健) |
作者单位 | 1.Snap Inc. 2.快手 3.京东 4.亚马逊 5.中国科学院大学 6.中科院自动化所 |
推荐引用方式 GB/T 7714 | Zhao TL,Zhang X,Zhu WT,et al. Multi-Granularity Pruning for Model Acceleration on Mobile Devices[C]. 见:. 线上. 2022-07. |
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