学术报告报告题目: Efficient Neural Architecture Search and its Applications in Computer Vision
报 告 人：董宣毅 博士(悉尼科技大学)
报告时间：2019年3月5日（星期二）下午16:10-17:10
报告地点：4号楼318会议室
邀 请 人：刘深泉 教授
欢迎广大师生前往！
数学学院
2019年3月4日
报告摘要：
Representation learning is a fundamental research problem in computer vision, because it benefits downstream computer vision applications, such as detection and segmentation. Due to the success of deep learning, representation learning has undergone a transition from “feature engineering” (SIFT/HOG) to “architecture engineering” (GoogleNet/VGG/ResNet). However, a lot of expert knowledge and ample computational resources are still required to secure a good architecture that can capture robust representations. This presentation will provide insight into algorithm design to discover robust architectures to save GPU resources and increase speed.
报告人简介：
Xuanyi Dong is a second-year Ph.D. student at University of Technology Sydney (UTS), under the supervision of Prof. Yi Yang. In 2016, he received the B.Eng. degree from Beihang University. In 2017, he worked at Facebook Reality Labs as a research intern. To date, he has published more than ten papers on CCF A conferences or CCF A journals, including TPAMI, TIP, and CVPR. He is a Programm Committee member of CVPR 2019, ICCV 2019, and ACM MM 2017-2018. He is also a reviewer for TPAMI, TIP, IJCV, TCSVT, etc. His research interests are automated machine learning and its applications in computer vision.