Zachary Lubberts

Hi there!

I'm a data scientist. In my case, that means: some optimization, some programming, some signal processing, some probability and statistics, and a lot of linear algebra. My recent work has focused on statistics on graphs, and natural language processing.

I finished my PhD at Johns Hopkins University in 2019, advised by Youngmi Hur (now at Yonsei University in Seoul, Korea). In my dissertation, Generating Tight Wavelet Frames from Sums of Squares Representations, I used tools from real algebraic geometry to construct nonseparable multivariate tight wavelet frames with many vanishing moments. These frames can be used to better capture signals with strong directional content that is not aligned with the main coordinate axes, useful for compression, denoising, or signal extraction. Before that, I studied Philosophy and Applied Math and Statistics in undergraduate, also at JHU (2010-2013).

I teach courses on a wide range of topics, including mathematical computing, optimization, probability and statistics, and real analysis, to students at all levels of undergraduate, masters, and PhD study, and in several different modalities. This spring, I'm teaching the graduate-level Statistical Machine Learning course (STAT 5630).

Recent Updates

• "Euclidean Mirrors and Dynamics in Network Time Series" is now available at the Journal of the American Statistical Association!

• "Curvature-based Clustering on Graphs" is now available on arXiv: arXiv:2307.10155 

• "Discovering a change point and piecewise linear structure in a time series of organoid networks via the iso-mirror" is now available on arXiv: arXiv: 2303.04871 

• "Joint Spectral Clustering in Multilayer Degree-Corrected Stochastic Blockmodels " is now available on arXiv: arXiv:2212.05053 

• "Discovering Underlying Dynamics in Time Series of Networks" has been updated on arXiv: arXiv:2205.06877

• "Mixed-Membership Community Detection via Line Graph Curvature" has been accepted as a poster at NeurReps 2022.

• "Entrywise Estimation of Singular Vectors of Low-Rank Matrices With Heteroskedasticity and Dependence" is now available at IEEE Transactions on Information Theory: DOI here arXiv:2105.13346 

• "Beyond the adjacency matrix: random line graphs and inference for networks with edge attributes" is now on arXiv: arXiv:2103.14726