Jonathan Stewart
Florida State University, FL·Mathematics & Statistics
Networks are everywhere data is collected on who interacts with whom: friendships, collaborations, disease contacts, protein interactions. Statisticians model such data to learn the hidden structure behind the observed links, but a standard shortcut sits inside most of these models.
- Fields
Mathematics & Statistics
Social Science
Computer Science
- Topics
network data
latent variable models
latent space models
- Grants
- 1 known grant, through May 2028
What they work on
- Develop a new class of generalized latent position models for network data that drop the usual conditional independence assumption, so that dependence among connections, not just node positions, shapes the modeled network.
- Integrate and extend exponential-family random graph models, which encode local dependence through features like shared partners, with latent space models, which place nodes in a hidden geometry, using each to offset the other's known deficiencies.
- Build estimation methodology for fitting the proposed models to observed networks, together with statistical theory establishing the foundations for parameter estimation and inference under edge dependence.
- Apply the new models to real network data from the social and life sciences and release freely distributed open-source software implementing them for practitioners.
More about the research
They assume that, once you know each node's hidden position or group, the presence of one connection tells you nothing about any other connection. Real social and biological networks violate this: people befriend friends of friends, triangles close, and popularity feeds on itself. This project asks how to build models that keep the strengths of hidden-position modeling while also representing these dependencies among connections.
Why it matters
Conclusions drawn from network data, about how behaviors or infections spread, how organizations function, or how biological molecules interact, depend on whether the model matches how the network actually formed. Models that wrongly assume connections are independent can mislead. By offering models that capture realistic dependence along with hidden structure, plus open-source software, this work could give researchers across the social and life sciences more trustworthy tools for analyzing relational data.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Carter Butts
University of California-Irvine, CA·Mathematics & Statistics
Advance network regression methods using kernel learning, which relies on functions that score the similarity between complex objects, to measure how attributes and geography shape social ties.
Weijing Tang
Carnegie Mellon University, PA·Mathematics & Statistics
Develop priority-aware rules for valuing training data based on the Shapley value, a game-theory method for fairly splitting credit, that account for originality, provenance and individual risk within a unified axiomatic framework.
Joshua Cape
University of Wisconsin-Madison, WI·Mathematics & Statistics·Early-career
Develop nonparametric spectral clustering and dimension-reduction methods that tolerate messy, mixed data, and apply them to brain network studies of schizophrenia with psychiatry collaborators.
The school
Florida State University: 15 labs on PI Pages, $461M a year on research, 481 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.