Generative Zero-shot Network Quantization | |
Xiangyu, He1,2; Jiahao, Lu1,2; Weixiang, Xu1,2; Qinghao, Hu2; Peisong, Wang2; Jian, Cheng2 | |
2021-06 | |
会议日期 | 2021-6 |
会议地点 | Virtual Event |
英文摘要 | Convolutional neural networks are able to learn realistic image priors from numerous training samples in low-level image generation and restoration [66]. We show that, for high-level image recognition tasks, we can further recon struct “realistic” images of each category by leveraging intrinsic Batch Normalization (BN) statistics without any training data. Inspired by the popular VAE/GAN method s, we regard the zero-shot optimization process of synthet ic images as generative modeling to match the distribution of BN statistics. The generated images serve as a calibra tion set for the following zero-shot network quantizations. Our method meets the needs for quantizing models based on sensitive information, e.g., due to privacy concerns, no data is available. Extensive experiments on benchmark datasets show that, with the help of generated data, our ap proach consistently outperforms existing data-free quanti zation methods. |
语种 | 英语 |
内容类型 | 会议论文 |
源URL | [http://ir.ia.ac.cn/handle/173211/48941] |
专题 | 类脑芯片与系统研究 |
作者单位 | 1.University of Chinese Academy of Scienses 2.Institute of Automation, Chinese Academy of Sciences |
推荐引用方式 GB/T 7714 | Xiangyu, He,Jiahao, Lu,Weixiang, Xu,et al. Generative Zero-shot Network Quantization[C]. 见:. Virtual Event. 2021-6. |
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