Dependence detection in high-dimensional regression problems

What We Do

Research problems concern discovery of a minimal set of predictors influencing the target (its Markov Blanket) , prediction interaction between features and modeling of causal dependencies in uplift. Studies use mainly non-parametric approaches based on information-theoretic measures and adaptation of data-mining techniques to uplift. One of the goasl is to
construct stopping rules for greedy search of active variables.



Key works

Recent contributions

  • J. Mielniczuk, P. Teisseyre (2019) Stopping rules for information-based feature selection, Neurocomputing, 255-274
  • K. Rudaś, , S. Jaroszewicz (2018) Linear regression for uplift modeling. Data Mining and Knowledge Discovery, 1275-1305.
  • J. Mielniczuk, P. Teisseyre (2018) A Deeper Look at Two Concepts of Measuring Gene-Gene Interactions: Logistic Regression and Interaction Information Revisited, Genetic Epidemiology
  • M. Kubkowski, J. Mielniczuk (2018) Projections of a general binary model on a logistic regression, Linear Algebra and its Applications, 536, 152-173
  • J. Mielniczuk, M. Rdzanowski (2017) Use of information measures and their approximations to detect predictive gene-gene interactions, Entropy, 19, 1-23

more publications


Jan Mielniczuk


Have a Project in Mind?

Please get in touch with the head of the group.