南京大学计算机科学与技术系软件新技术与产业化协同创新中心
摘 要：
In the past few years, progresses have been shown that much of the visual problems (e.g. classification, detection, segmentation) can be approached by collecting a large amount labeled data, and training a huge neural net. This draws scalability limitations to the open world where new objects may constantly appear, and labeling costs are often hard to manage.
I will show our recent progress on recognition with unsupervised learning and learning with scarce annotations. Our key insights are two folds: 1) A simple unsupervised learning algorithm by discriminating instances could achieve the state-of-the-art performance. 2) Advances in unsupervised learning can directly translate to few-shot recognition. 3) Simple label propagation algorithms can create abundance of labeled data reliably by propagating labels to the unlabeled data. We demonstrate state-of-the-art performance for several image and video recognition problems under the constraints that labeled data is scarce.
报告人简介:
Zhirong Wu is currently a Researcher in the visual computing group at Microsoft Research Asia. Previously, he obtained his B.Eng from the department of automation at Tsinghua University, and his Ph.D from the Chinese University of Hong Kong advised by Prof. Xiaoou Tang. He was a visiting PhD student at Princeton Vision Group advised by Prof. Jianxiong Xiao. He was Post-doctoral scholar at UC Berkeley with Dr. Stella Yu.
时间：9月25日(星期三）9:00
地点：计算机科学技术楼222室
|
