Stanford CS224W: Machine Learning with Graphs | 2021 | Lecture 3.3 - Embedding Entire Graphs

For more information about Stanford’s Artificial Intelligence professional and graduate programs, visit: https://stanford.io/316zi1Z Jure Leskovec Computer Science, PhD In some scenarios it is important to not only learn embeddings for nodes, but also the entire graph. In this video, we introduce several approaches that could effectively learn embeddings for entire graphs, including aggregation of node embeddings, as well as the anonymous walk embedding approach. To follow along with the course schedule and syllabus, visit: http://web.stanford.edu/class/cs224w/ 0:00 Introduction 0:27 Embedding Entire Graphs 0:58 Approach 2 2:54 Approach 3: Anonymous Walk Embeddings 5:02 Number of Walks Grows 5:52 Simple Use of Anonymous Walks 7:09 Sampling Anonymous Walks 8:27 New idea: Learn Walk Embeddings 13:59 Preview: Hierarchical Embeddings 14:32 How to Use Embeddings 16:21 Today's Summary