컨텐츠 시작
학술대회/행사
초록검색
| 제출번호(No.) | 0190 |
|---|---|
| 분류(Section) | Poster Session |
| 분과(Session) | (AM) Applied Mathematics(including AI, Data Science) (AM) |
| 발표시간(Time) | 19th-B-14:00 -- 14:30 |
| 영문제목 (Title(Eng.)) |
Recent studies about Graph Neural Diffusion framework |
| 저자(Author(s)) |
Hyunjun Kim1 KAIST1 |
| 초록본문(Abstract) | Over recent years, significant advancements have been made in machine learning, particularly in the domain of neural network architectures. This progress has enabled AI models to not only process Euclidean data but also non-Euclidean data, notably graphs, which are ubiquitous in diverse fields such as science and social networks. This poster provides an overview of graph neural networks (GNNs) along with specific models that have emerged in this domain. The Graph Convolution Network (GCN), which leverages neighborhood concepts, and the Graph Attention Network (GAT), which enhances GCN through attention mechanisms, are discussed. Additionally, the neural ordinary differential equation (neural ODE) model is introduced, offering a novel approach to interpreting feedforward networks as ODE models, thereby improving computational efficiency. Finally, we introduce the Graph Neural Diffusion (GRAND) framework, which treats GNNs as discretizations of underlying Partial Differential Equations (PDEs). |
| 분류기호 (MSC number(s)) |
68T07 |
| 키워드(Keyword(s)) | Graph Neural Network, GRAND |
| 강연 형태 (Language of Session (Talk)) |
Korean |