수 학 과

일 시 : 2024년 04월 16일(화) 16:30~17:30 

연 사 : 우영호 (국가수리과학연구소)

제 목 : Topological vector representation of graph and graph neural network

장 소 : 자1204호

연 락 : 주항연 교수

주 최 : 충남대학교 수학과, 수리과학연구소

초록: In this presentation, we delve into various methods for representing graphs—mathematical constructs that encode a wide array of non-Euclidean data, including molecules, social networks, genes, transportation networks, and citation networks, among others. The conversion of graph data into Euclidean vector is a critical step in applying machine learning techniques for graph classification, a process that is intricately linked to the field of graph neural networks in computer science. We will explore several key methodologies, including the Weisfeiler-Lehman algorithm, random walks, and graph convolution, which are widely utilized techniques in this area. Additionally, we will discuss the outcomes of integrating these methods with topological invariants of graphs, offering insights into their efficacy and application in analysis of relational data