讲座题目：《Research on Neural Network based Graph Representation Learning Methods and Applications》

主讲嘉宾：权沛讲师

讲座时间：2023年10月31日（星期二）16:00-17:30

讲座地点：沙河校区/南路校区 具体地点

讲座摘要：

Graphs are a complex but ubiquitous data structure. The data of many problems in the real world can be described with graphs, such as the traffic network extending in all directions, the protein graph that is closely related to our health, and the social network that describes the social relationship between people. In addition to the feature information in common data, graph data also have edges that describe the relationship between nodes. It not only endowsthe graph with more powerful expression ability but also poses challenges for its representation, making it difficult for many graph-related tasks to achieve better results. Since the outstanding achievements of neural networks in many fields, these architectures are generalized to the graph domain to generate comprehensive representation from structure and feature information. However, due to the characteristics of non-regular structures and unique edge information, the research on graph neural networks is still in its infancy, and the utilization of graph information is still not enough. Based on the above background, this work will explore the determination of information sources by considering the characteristics of graph data, and research the representation and application of graph data based on neural networks.嘉宾简介：权沛，计算机应用技术工学博士，现任北京工业大学经济与管理学院管理科学与工程系讲师，Annals of Data Science领域编委，国际信息技术与量化管理大会（ITQM）组委会委员。主要研究方向为图表示学习方法及应用、数据挖掘与深度学习理论、自然语言处理等，现主持北京市博士后科研活动资助项目1项。近五年在IEEE Transactions on Cybernetics、European Journal of Operational Research、Neural Networks等国际知名期刊与会议发表学术论文20余篇。曾获2022年中国科学院虚拟经济与数据科学研究中心优秀毕业生、2022年中国科学院大学优秀员工等多项奖励。

主办单位：12BET