报告题目：CNN Architecture Design: from Deeper to Wider

报告人：王井东 研究员

单位：微软亚洲研究院

报告时间：2017年5月19日（周五）上午9:15

报告地点：逸夫楼408会议室

报告摘要：In this talk, I will introduce a deep neural network design pattern, deep fusion. The resulting network, called deeply-fused net, is a multi-path, multi-scale, weight shared net with express ways between layers. Then, I present two deep fusion methods with improved parameter and computation efficiency. One is based on merge-and-run mappings, which improve the ResNets by going wider but less deep. The other one is primal-dual group convolution (PDGC). It is more efficient in parameter and computation than Xception and regular convolutions, which, we show, are special instances of our approach PDGC. I point out that going wider is another direction for CNN architecture design.

报告人简介：

Jingdong Wang is a Lead Researcher at the Internet Media Group, Microsoft Research Asia. His areas of interest include computer vision, multimedia, and machine learning. At present, he is mainly working on deep learning, human understanding, person re-identification, image recognition, and indexing and compact coding for large scale similarity search. He has published 100+ papers in top conferences and prestigious international journals, such as CVPR, ICCV, ACMMM, ICML, SIGIR, TPAMI, IJCV, and so on, and one book. His paper was selected into the best paper finalist at ACMMM 2015. He has shipped a dozen of technologies to Microsoft products, including Bing image search, Cognitive service, and XiaoIce Chatbot.

He has served/will serve as an area editor for TMM, an area chair in AAAI 2018, ICCV 2017, ICIP 2017, CVPR 2017, ECCV 2016, ACMMM 2015 and ICME 2015, a track chair in ICME 2012, a special session chair in ICMR 2014, and a program committee member or a reviewer in top conferences and journals, including CVPR, ICCV, ACMMM, NIPS, SIGIR, SIGGRAPH, TPAMI, IJCV.

太阳集团tyc5997