学术交流

新葡萄娱乐官网学术报告 第2020-12-08期

发布时间:2020-12-08动态浏览次数:15

Title

Making Sense of Social Media Data

  —Learning from sparse, high-dimensional, and privacy-sensitive data

Lecture & Organization

Prof. Dingqi Yang, University of Macau

Date & Time

8:00-9:30pm Tue, Dec 08, 2020

Beijing time

Mode

Zoom ID:  4046260856

Password: 000321

Bio

Dingqi Yang is currently an Associate Professor in the Department of Computer and Information Science, Faculty of science and technology at the University of Macau. Before that, he was a senior researcher in the eXascale Infolab at the University of Fribourg, Switzerland from May 2015 to Oct. 2020. He received his Ph.D. (with highest honors) in Computer Science from Pierre and Marie Curie University (Paris VI) and Institut Mines-Telecom/Telecom SudParis in Jan. 2015, where he won both the French National Centre for Scientific Research (CNRS) Samovar Doctorate Award ("Prix Doctorant CNRS Samovar/Télécom SudParis") and Institut Mines-Télécom Press Mention. He is broadly interested in designing novel data mining and machine learning techniques for urban Big Data analytics (in particular using large-scale social network data, spatiotemporal data, and graph data), and also in building practical systems for real-world smart city applications. More details can be found on his homepage (https://sites.google.com/site/yangdingqi/).

Abstract

Social media serves as an important information source for every internet user nowadays. Active social media users represent 49% of the world total population by Jan. 2020, generating a tremendous volume of data on the Web. On one hand, such user-generated data provides unprecedented opportunities for building intelligent Web applications. On the other hand, such social media data is intrinsically heterogeneous, sparse, high-dimensional, dynamic, and last but not least, privacy-sensitive. In this talk, I will introduce our practice in making sense from social media, including 1) predictive modeling of heterogeneous and sparse data, 2) representation learning for high-dimensional and dynamic data, and 3) Privacy protection of user data.