题目：Unifying Structural Proximity and Equivalence for Network Embedding
报告人：史本云，南京工业大学计算机科学与技术学院
摘要：The fundamental purpose of network embedding is to automatically encode each node in a network as a low-dimensional vector, while at the same time preserving certain characteristics of the network. Based on the nodes’ embeddings, downstream network analytic tasks like community mining, node classification, and link prediction, can then be easily implemented using traditional machine learning methods. In recent years, extensive network embedding methods have been proposed based on matrix factorization, random walks, deep learning, and so on. However, most of them focus mainly on preserving the structural proximity of network nodes, where highly interconnected nodes in a network will be represented closely together in the embedded vector space. On the other hand, existing studies on many real-world networks have revealed that high-order organizations, such as network motifs and graphlets, may be related to specific network functions. In this case, nodes far apart but with a similar organization in a network (i.e., structural equivalence) may have similar network functions.
In this talk, we will present two hybrid node embedding methods based on matrix factorization and random walks, respectively, which unify both structural proximity and equivalence of a network. Specifically, we adopt the concept of graphlet degree vector (GDV) to measure structural equivalence between network nodes. For the factorization-based method, we present a projection scheme and use one hyperparameter to balance the degree of importance between structural proximity and equivalence. For the random walks-based methods, we present a hybrid embedding method via cross-layer random walks in multiplex networks. Through carrying out experiments on both synthetic and real-world datasets, we evaluate the performance of the proposed hybrid embedding methods in tasks of node clustering, node classification, and visualization. The results demonstrate that the proposed methods outperform several state-of-the-art methods when the network analytic tasks are not merely related to structural proximity.
In addition to structural properties, it is also important and essential to preserve the dynamic properties of epidemic spreading on complex networks. Inspired by random walks-based methods, in this talk, we will also introduce an embedding method, namely EpiRep, to learn node representations through simulating epidemic dynamics on networks, by maximizing the likelihood of preserving groups of infected nodes due to the epidemics starting from every single node. The proposed method and findings in this paper may offer new insight for source identification and infection prevention in the face of epidemic spreading on social networks.
报告人信息：Benyun SHI (史本云) is currently a professor with the School of Computer Science and Technology, Nanjing Tech University, China. He received his B.Sc. degree in Mathematics from Hohai University, Nanjing, China, in 2003, and the M.Phil and Ph.D. degrees in Computer Science from Hong Kong Baptist University, Hong Kong, in 2008 and 2012, respectively. In the past, he also worked as a full professor at Hangzhou Dianzi University, China, and a Research Assistant Professor at Hong Kong Baptist University, Hong Kong. His research interests are in the areas of Data-driven Modeling and Analytics, Machine Learning, Complex Systems/Networks, Multi-Agent Autonomy-Oriented Computing, Computational Epidemiology and Health Informatics.