Research Interests
I work mainly in the field of functional data analysis (FDA), which
is a sub-field of mathematical statistics, centered around the problem
of statistical inference on the law of a continuous-time random process
given multiple realizations of such a process. I am particularly
interested in functional data on multi-dimensional domains, where the
random process is not only a function of time but also e.g. of space. My
research focuses on methodological aspects, but computational costs and
numerical aspects naturally enter into consideration due to the size and
continuity of multi-dimensional functional data. In another line of
research, I develop optimization algorithms for NP-hard versions of
principal component analysis (PCA) such as matrix completion, robust PCA
or sparse PCA. The intersection between my two main lines of research is
the concept of low-rankness (in different forms), which I consider to be
powerful and useful in many applied problems.
Work in Progress
- K. Waghmare, T. Masak & V. Panaretos: Functional Graphical
LASSO
- C. Kuemmerle, T. Masak & F. Krahmer: Optimal Quadratic Models
for Low-rank Optimization
Publications
- T. Masak & V. Panaretos (2022) Random Surface Covariance
Estimation by Shifted Partial Tracing. Journal of the American
Statistical Association.
- T. Masak, S. Sarkar & V. Panaretos (2023) Separable Expansions
for Covariance Estimation via the Partial Inner Product.
Biometrika.
- T. Masak, T. Rubín & V. Panaretos (2022) Sprsely Observed Random
Surfaces. Journal of Computational and Graphical Statistics.
- T. Masak (2017) Iteratively Reweighted Least Squares Algorithm for
Sparse Principal Component Anlysis with Application to Voting Records.
Statistika: Statistics and Economy Journal.
Software
surfcov
package
- publicly
avilable
- covariance estimation for random surfaces beyond separability
- implements methodologies from Masak & Panaretos (2022) and
Masak, Sarkar & Panaretos (2023)
cerss
package
- publicly
available
- allows reproduction of the simulation studies and real data
applications in Masak & Panaretos (2022) and Masak, Sarkar &
Panaretos (2023) and Masak, Rubín & Panaretos (2022)