正文
个人主页：
https://cs.adelaide.edu.au/~chhshen/
报告摘要：
The talk will cover Prof. Shen’s recent works on NAS for semantic segmentation, object detection and image restoration.
参考文献：
[1]
Fast Neural Architecture Search of Compact Semantic Segmentation Models via Auxiliary Cells，Vladimir Nekrasov, Hao Chen, Chunhua Shen, Ian Reid，CVPR 2019，https://arxiv.org/abs/1810.10804.
[2]
NAS-FCOS: Fast Neural Architecture Search for Object Detection, Ning Wang, Yang Gao, Hao Chen, Peng Wang, Zhi Tian, Chunhua Shen, https://arxiv.org/abs/1906.04423.
[3]
IR-NAS: Neural Architecture Search for Image Restoration，Haokui Zhang, Ying Li, Hao Chen, Chunhua Shen，https://arxiv.org/abs/1909.08228.
报告嘉宾：
闫俊杰（商汤科技）
报告时间：
2019年10月9日（星期三）晚上19:30（北京时间）
报告题目：
AutoML in Full Life Circle of Deep Learning Assembly Line
报告人简介：
Junjie Yan is the CTO of Smart City Business Group and Vice Head of Research at SenseTime. He leads the R&D Team within Smart City Group to build systems and algorithms that make cities safer and more efficient. At SenseTime Research, he leads the Deep Learning Toolchian Team to build deep learning toolchain from algorithm components to distributed training and inference platform that scale up deep learning solutions to more than 700 customers. His research interests include deep learning and system.
个人主页：
http://yan-junjie.github.io
报告摘要：
This talk introduces a series of papers and best practices from SenseTime that aim to build deep learning assembly line with AutoML techniques in full life circle, including network architecture search, loss, data augmentation, data sampling, and model deployment.
参考文献：
[1]
Efficient Neural Architecture Transformation Searchin Channel-Level for Object Detection, Junran Peng, Ming Sun, Zhaoxiang Zhang, Tieniu Tan, Junjie Yan, NeurIPS 2019, https://arxiv.org/abs/1909.02293.
