Towards Convolutional Neural Networks Compression via Global&Progressive Product Quantization | |
Chen, Weihan1,2; Wang, Peisong1,2; Cheng, Jian1,2 | |
2020 | |
会议日期 | 2020 |
会议地点 | Online |
英文摘要 | In recent years, we have witnessed the great success of convolutional neural networks in a wide range of visual applications. However, these networks are typically deficient due to the high cost in storage and computation, which prohibits their further extensions to resource-limited applications. In this paper, we introduce Global&Progressive Product Quantization (G&P PQ), an end-to-end product quantization based network compression method, to merge the separate quantization and finetuning process into a consistent training framework. Compared to existing two-stage methods, we avoid the timeconsuming process of choosing layer-wise finetuning hyperparameters and also make the network capable of learning complex dependencies among layers by quantizing globally and progressively. To validate the effectiveness, we benchmark G&P PQ by applying it to ResNet-like architectures for image classification and demonstrate state-of-the-art tradeoff in terms of model size vs. accuracy under extensive compression configurations compared to previous methods. |
内容类型 | 会议论文 |
源URL | [http://ir.ia.ac.cn/handle/173211/40618] |
专题 | 自动化研究所_模式识别国家重点实验室_图像与视频分析团队 |
通讯作者 | Cheng, Jian |
作者单位 | 1.University of Chinese Academy of Sciences 2.Institute of Automation, Chinese Academy of Sciences |
推荐引用方式 GB/T 7714 | Chen, Weihan,Wang, Peisong,Cheng, Jian. Towards Convolutional Neural Networks Compression via Global&Progressive Product Quantization[C]. 见:. Online. 2020. |
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