Welcome!

We are the Graph Exploration and Mining at Scale (GEMS) lab at the University of Michigan, founded and led by Danai Koutra. Our team researches important data mining and machine learning problems involving interconnected data: in other words, graphs or networks.

From airline flights to traffic routing to neuronal interactions in the brain, graphs are ubiquitous in the real world. Their properties and complexities have long been studied in fields ranging from mathematics to the social sciences. However, many pressing problems involving graph data are still open. One well-known problem is scalability. With continual advances in data generation and storage capabilities, the size of graph datasets has dramatically increased, making scalable graph methods indispensable. Another is the changing nature of data. Real graphs are almost always dynamic, evolving over time. Finally, many important problems in the social and biological sciences involve analyzing not one but multiple networks.

So, what do we do?

The problems described above call for principled, practical, and highly scalable graph mining methods, both theoretical and application-oriented. As such, our work connects to fields like linear algebra, distributed systems, deep learning, and even neuroscience. Some of our ongoing projects include:

• Algorithms for multi-network tasks, like matching nodes across networks
• Learning low-dimensional representations of networks in metric spaces
• Abstracting or “summarizing” a graph with a smaller network
• Analyzing network models of the brain derived from fMRI scans
• Distributed graph methods for iteratively solving linear systems
• Network-theoretical user modeling for various data science applications

We’re grateful for funding from Adobe, Amazon, the Army Research Lab, the Michigan Institute for Data Science (MIDAS), Microsoft Azure, the National Science Foundation (NSF), and Trove.

Interested?

If you’re interested in joining our group, send an email with your interests and CV to gemslab-opportunities@umich.edu.