Subject of interest : Community detection & random graphs.
My researches focus on three areas:
- Clustering in multiplex and temporal networks. Our preprint Estimation of Static Community Memberships from Temporal Network Data (arXiv preprint @ arXiv:2008.04790) establishes some information-theoretic bounds for community recovery, and proposes a likelihood-based, online algorithm for clustering temporal networks.
- Clustering in geometric graphs: our publication Higher-Order Spectral Clustering for Geometric Graphs @ Journal of Fourier Analysis and Applications shows that when clustering geometric graphs, the second eigenvector of the graph Laplacian (so-called Fiedler vector) might not be meaningful, and one has to look at higher-order eigenvector.
- Semi-supervised graph clustering: we have established some bounds on the recovery error made by semi-supervised extensions of spectral methods. One preprint @ arXiv:2007.14717 and one conference publication Almost Exact Recovery in Label Spreading @ International Workshop on Algorithms and Models for the Web-Graph.
I am a member of the NEO team @ Inria Sophia-Antipolis (France); my PhD is under the supervision of Konstantin Avrachenkov. I have also worked with Lasse Leskelä from Aalto University (Finland).