Network science
Many social and biological network systems are characterized by highly sophisticated interactions and dynamics between agents. These networks often exhibit high sparsity, severe degree heterogeneity and mixed memberships. One task of great scientific interest is to identify the network’s latent community structure. My research has focused on problems including global testing in static networks and mixed membership estimation in dynamic networks.
Article
Cammarata, L.V. and Ke, Z.T., 2023. Power Enhancement and Phase Transitions for Global Testing of the Mixed Membership Stochastic Block Model. Bernoulli.
In this work, I derived the phase transition of global detection (i.e., testing whether a network has more than one community) in the Mixed Membership Stochastic Block Model and I designed a novel statistically optimal and computationally tractable hypothesis test called the Power-Enhancement Test.
Manuscript in preparation
Cammarata, L.V., Jin, J. and Ke, Z.T., 2024. Dynamic network analysis under smooth community evolvement and potentially erratic degree changes. Manuscript in preparation.
In this work, I develop a rate optimal dynamic mixed-membership estimation and embedding algorithm using spectral methods in the degree-corrected mixed membership stochastic block model. I apply the algorithm to international trade networks and dynamic human contact networks.