Da is an applied scientist at Amazon. He got a PhD in computer science from the Johns Hopkins University. His research covers large-scale machine learning and data-intensive computing. His current work focuses on the research and development in MXNet and DGL.
All real-world data has structures that are best described as graphs. If there is one data structure for deep learning algorithms, graph would be the foremost candidate. The graph structure can be either explicit, such in social networks, knowledge graphs, and protein-interaction networks, etc., or latent and implicit, as in the case of languages and images. Leveraging and discovering graph structures have many immediate applications and also serves as a futile ground for the next generation of algorithms.
In this talk, we will first introduce graph neural networks in general. We will then discuss their business applications such as personalization and recommendation, and touch upon some interesting research directions. We will demo how to use DGL's API to develop the popular graph convolution network (GCN) and scale it to very large graphs in the EC2 cloud. We will also discuss popular training algorithms for GCN on giant graphs and the best practice of using these algorithms in the EC2 cloud.
新丸ビル, Room 902
2019年04月24日 午後04:30 - 午後05:00