主办单位：计算机与信息工程学院、国际合作与交流处

报告时间：2021年11月26日（周五），15：30

报告地点：腾讯会议ID: 615 680 658

报告人： Fadi Dornaika（河南大学外籍拔尖人才特聘教授）

  报告摘要：Learning with graph data has attracted increasing attention recently. Graph neural network (GNN) has become the main tool for learning representations on graphs, which is an extension of traditional neural networks for processing graph data. GNN processes data and its graphs together, with the main goal of creating a new representation of the data and solving many machine learning problems.This talk contains an introduction to general graphs used in computer science and operational research. Then, some principles of GNN will be introduced. In particular, the GCN, a widely used type of GNN, is introduced from both spectral theory and spatial perspectives.GraphSage and the graph attention model, as well as some variants of GCNs, will also be introduced. Notions such as graph pooling and graph attacks are briefly introduced.

  报告人简介: Fadi Dornaika is currently a professor of the University of the Basque Country UPV/EHU, San Sebastian, Spain, and a research professor at Ikerbasque, Basque Foundation for Science, Bilbao, Spain. He has also been appointed as a distinguished visiting professor of Henan University, China. He received the Ph.D. in Computer Science from INRIA and the Grenoble Institute of Technology, Grenoble, France, in 1995. In the past, he held various research positions in Europe and Canada. He has published more than 300 papers in the fields of computer vision, pattern recognition and machine learning, including many top conference and journal papers. His current research interests include Semi-Supervised Learning and Manifold Learning, Deep Learning with applications to facial age estimation, facial beauty prediction, emotion recognition, fatigue detection, etc.

  报告人简介： Fadi Dornaika是西班牙国立巴斯克大学，计算机与人工智能系教授，及河南大学计算机与信息工程学院，外籍拔尖人才特聘教授。1995年博士毕业于INRIA及Grenoble Institute of Technology, France.  之后在欧洲和加拿大的多所科研机构任职。发表论文300余篇，含多篇顶会顶刊论文。主要研究兴趣是人工智能和模式识别，含半监督学习、流形学习、深度学习及应用（人脸属性识别、表情识别、年龄估计等）、及图神经网络等细分方向。