My research covers a broad range of statistical learning and advanced machine learning methods that are needed in modern data analysis problems. The aim is to take into account for a) complexity of the data including latent low rank structures and subspaces, sparsity and missing values, or the sheer variety of the data, b) large scale settings which refers to high-dimensionality but also settings where the sample size is smaller or not much larger than the data dimension which make traditional asymptotically optimal methods perform poorly and c) dynamic nature of the data, or data velocity, where data accumulates or streams at fast pace and thus batch processing is not a viable solution.

A particular focus area is the analysis of high-dimensional (HD) sparse data, i.e., data for which the sample size n is smaller or not much larger than the dimension p of the data set, where p is potentially very large. In such cases, the number of parameters to estimate can greatly exceed the number of observations. For example, in genomic studies, n is often the number of patients (only few tens!) and p is the number of genes (tens of thousands). Yet a classifier needs to learn a high-dimensional parameter with a limited data. This demands new approaches such as regularized optimization or imposing some structure on the unknown parameter to reduce the number of unknowns. Sparse HD data sets are becoming more common place in practice, particularly with the development of areas such as genomics, multimedia imaging, or financial economics.

Doctoral students

  • Xinjue Wang (6/2022 - )
  • Lei Wang (3/2024 - )

    • Research assistants

  • Juho Kuikka

    • Visitors

  • Leatile Marata (8/2023-2/2024)

    • Alumni

    raninen Elias Raninen, engineer at Nokia Bell Labs
    Doctoral student 6/2017- 6/2022. Thesis: Contributions to the Theory and Estimation of High-dimensional Covariance Matrices
    M.Sc. student (6/2016-5/2017). Thesis: Scaled sparse linear regression with the elastic net

    tabassum Muhammad Naveed Tabassum, R&D and Specification Engineer at Nokia
    Doctoral student 1/2016 - 3/2021.
    Thesis: Sparsity Driven Statistical Learning for High-Dimensional Regression and Classification

    ouzir Nora Ouzir, Assistant professor at CentraleSupelec, France.
    post-doc 11/2018 - 12/2020 (jointly supervised with Prof. S. Vorobyov)
    Topic: ultrasound imaging, dictionary learning

    miettinen Jari Miettinen
    post-doc 8/2017 - 7/2020 (jointly supervised with Prof. S. Vorobyov)
    Topic: Graph signal processing

    ammar Ammar Mian, Associate professor at University Savoie Mont Blanc, France
    post-doc 10/2019 - 9/2020
    Topic: Machine learning methods on manifolds, pedestrian detection.

    basiri
     Shahab Basiri, Reserach Scientist at Varian Medical Systems
    Doctoral student 3/2014 - 6/2018.
    Thesis: Robust large-scale statistical inference and ICA using bootstrapping
    M.Sc. student 7/2012 - 2/2014.
    Thesis: Hypothesis Testing in Independent Component Analysis

    ejaz Aqib Ejaz, Senior Engineer at Emberion Oy
    M.Sc. student, 12/2014 - 5/2015.
    Thesis: Algorithms for Sparse Signal Recovery in Compressed Sensing

    ejaz Alireza Razavi, Senior AI Development Engineer at Scania Group
    Post-doctoral fellow, 9/2011-8/2012. Topic: Compressed sensing.