Graph Neural Networks: What’s behind the Hype?

How does AI predict traffic, recommend friends, and discover new drugs? Learn how Graph Neural Networks derive insights from complex, unstructured data.

#1about 2 minutes

Why graph neural networks excel with unstructured data

Graph neural networks are uniquely suited for unstructured data like 3D meshes and social networks where traditional CNNs struggle.

#2about 3 minutes

Reviewing core concepts from graph theory

Key graph theory concepts are explained, including nodes, edges, directed vs undirected graphs, and homogeneous vs heterogeneous graphs.

#3about 2 minutes

Choosing the right data structure for graphs

An adjacency matrix is suitable for small graphs, while an adjacency list is more spatially efficient for large, sparse graphs.

#4about 3 minutes

A brief refresher on deep learning fundamentals

The core deep learning process of training and inference is reviewed, along with the distinction between supervised and unsupervised learning.

#5about 6 minutes

Exploring graph, node, and edge level prediction tasks

GNNs can perform predictions at the graph level (molecule properties), node level (community detection), and edge level (recommendation systems).

#6about 4 minutes

Understanding the GNN training and data splitting process

GNNs are trained using the message passing algorithm to create node embeddings, followed by a transductive split for training and validation sets.

#7about 2 minutes

Frameworks and resources for building GNNs

Popular frameworks like DGL, PyTorch Geometric, and TensorFlow GNN simplify the implementation of graph neural networks.

#8about 1 minute

Summary of key concepts in graph neural networks

The talk concludes with a recap of key takeaways, including graph modeling, data representation, prediction tasks, and the message passing algorithm.

#9about 3 minutes

Q&A on data leakage, knowledge graphs, and embeddings

The Q&A session addresses audience questions about data leakage in transductive splits, applying GNNs to semantic knowledge graphs, and comparing graph embeddings to word embeddings.