Removing Unwanted Variation: Some of my postdoctoral work on unifying and generalizing confounder adjustment methods.
Higher-order Spectral Estimators: My work on adaptive higher-order spectral estimation. These estimators generalize a class of popular matrix estimators to tensors. These estimators shrink tensors toward low multilinear-rank estimates.
The Incredible HOLQ: The incredible higher-order LQ decomposition for tensor data. This generalizes the matrix LQ decomposition to tensors and introduces the concept of a scaled all-orthonormal tensor.
Equivariant Estimation: A description of my dissertational work on optimal equivariant covariance estimation for tensor datasets.