Professor Moses Boudourides , from the Graduate Program on Data Science Northwestern University

In this tutorial, we will explore hypergraphs and how to leverage the Pythonic library PyTorch for learning from them, with a focus on making these concepts accessible to everyone, regardless of their background in computer science. Unlike traditional graphs, hypergraphs allow edges to connect any number of nodes, making them ideal for modeling complex group interactions that arise naturally in social networks, biological systems, and collaborative environments.

The session will take the form of a hands-on workshop. Participants will receive the Jupyter notebook in advance and can follow along on their laptops, running the code simultaneously with the presenter. No prior knowledge of Python is required, though participants should have Jupyter installed on their devices.

By the end of the tutorial, participants will understand what hypergraphs are and why they are important for representing multi-way relationships, learn how to represent and manipulate hypergraphs in Python using modern libraries, see how PyTorch enables machine learning on hypergraph structures through tensor operations and neural network architectures, and gain an introduction to the CHESHIRE algorithm, a state-of-the-art deep learning method for hyperlink prediction that employs Chebyshev spectral convolution to efficiently predict missing connections in complex networks and hypernetworks, including social and biological systems.

The tutorial emphasizes intuition and practical implementation, providing hands-on toy examples drawn from real-world applications. Participants will leave with a clear understanding of both the theory and practice of hypergraph learning.