Sreekalyani Bhamidi
University of North Carolina at Chapel Hill, NC·Mathematics & Statistics
Networks are everywhere in modern data: how information and misinformation spread through social media, how brain regions connect, how genes influence one another. Researchers increasingly feed network representations into multi-step machine learning pipelines and draw scientific conclusions from the output.
- Fields
Mathematics & Statistics
Computer Science
- Topics
network analysis
multilayer networks
functional data analysis
- Grants
- 1 known grant, through Aug 2028
What they work on
- Develop statistically principled methods for summarizing, clustering, and extracting principal directions of variation in multilayer network data, using Gaussian process optimal transport and Procrustes metrics on covariance operators expressed via Wasserstein distance.
- Analyze optimal-transport methods for time registration and supervised learning in evolving complex systems, motivated by single-cell RNA-seq developmental trajectories and network neuroscience, including clustering after quotienting out the trajectory.
- Study thresholding pipelines applied to covariance and correlation matrices in Gaussian graphical models through local weak convergence to limiting infinite structures, and derive implications for how thresholds should be selected.
- Connect hierarchical representation learning to continuum scaling limits via the relationship between linkage clustering and thresholding, for high-dimensional data with a latent network linking features.
- Characterize penalized estimation and limiting distributions of random adjacency matrices, including eigenvector localization, and use these results for hub detection in gene-expression data from cancer genomics.
More about the research
What is often missing is a theory telling us whether those conclusions are reliable. This project builds the mathematical foundations for two widely used families of network and high-dimensional methods, so users can know what the methods actually estimate and how accurate they are.
Why it matters
Conclusions about brain connectivity, cell development, and gene interaction networks increasingly rest on network and graphical-model pipelines whose statistical accuracy is unverified. Theory on convergence, thresholding choices, and hub detection would let analysts in neuroscience, single-cell biology, and cancer genomics choose tuning parameters defensibly and know when an apparent structure in their data is real rather than an artifact of the method.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Marianna Pensky
University of Central Florida, FL·Mathematics & Statistics
Extend the multiplex Generalized Dot Product Graph model so that groups of network layers can have different underlying structures. This would allow layers to be clustered by condition.
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.
Adrian Lewis
Cornell University, NY·Mathematics & Statistics
Extend Łojasiewicz-type rescaling techniques to analyze the complexity of first-order and active-set algorithms on semi-algebraic problems in hyperbolic, non-Euclidean and infinite-dimensional spaces.
The school
University of North Carolina at Chapel Hill: 72 labs on PI Pages, $1.6B a year on research, 523 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.