Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore the computational, algorithmic, and modeling challenges of analyzing massive graphs through this comprehensive Stanford University course taught by Jure Leskovec. Master machine learning techniques and data mining tools designed to extract insights from complex networks representing social, technological, and biological systems. Begin with fundamental concepts including graph representation choices and traditional feature-based methods for nodes, links, and entire graphs. Dive deep into node embeddings using random walk approaches and learn to embed entire graphs effectively. Study essential algorithms like PageRank, random walks with restarts, and matrix factorization techniques for node embeddings. Develop expertise in Graph Neural Networks (GNNs) through detailed exploration of their architecture, including single layers, stacking multiple layers, and designing maximally expressive GNN models. Learn advanced techniques for graph augmentation, training strategies, and prediction task setup for GNNs. Examine specialized applications including label propagation on graphs, heterogeneous graph machine learning, and knowledge graph embeddings with reasoning capabilities. Master query answering in knowledge graphs using Query2box methodology and explore neural subgraph matching and counting techniques. Understand community detection algorithms including the Louvain algorithm and overlapping community detection methods. Study GNN applications in recommender systems and explore deep generative models for graphs, including Graph RNNs for realistic graph generation. Address scalability challenges through techniques like GraphSAGE neighbor sampling, Cluster GCN, and simplified GNN architectures. Conclude with cutting-edge topics in geometric graph learning and trustworthy graph AI, gaining practical experience with influence maximization, disease outbreak detection, and comprehensive social network analysis throughout the curriculum.
