Graph neural networks (GNNs) are models designed to perform inference on unstructured data described by graphs. For different segments and industries, GNNs find suitable applications such as molecular analysis, drug discovery and repurposing, predicting stock market developments, thermodynamics, and even modeling human brain connectomes. Unlike conventional convolutional neural nets, GNNs address the challenge of working with data in irregular domains. We'll describe several ways to model unstructured problems as classification or regression at various levels, including graph-level, node-level, edge-level, graph-to-graph level, and graph learning. We'll also introduce concepts such as message passing on graphs, graph convolution with spatial filtering, and multilayer GNNs. Finally, we'll guide you through the process of creating your first GNN-based classifier using the Deep Graph Library, and accelerating it on NVIDIA GPUs with libraries like cuGraph. Prerequisite(s): Experience with deep learning or machine learning.
*Please disregard any reference to "Event Code" for access to training materials. "Event Codes" are only valid during the original live session.
Explore more training options offered by the NVIDIA Deep Learning Institute (DLI). Choose from an extensive catalog of self-paced, online courses or instructor-led virtual workshops to help you develop key skills in AI, HPC, graphics & simulation, and more.