Projects

Kristína Mezeiová

Project selection:

National projects

CAUSMET – Methods and algorithms for causal analysis and quantification of measurement uncertainty
Názov projektu Metódy a algoritmy kauzálnej analýzy a kvantifikácie neistôt meraní
Program: SRDA
Duration: 1.9.2026 – 31.12.2029
Project leader: Doc. RNDr. Witkovský Viktor, CSc.
Annotation: The project develops advanced methods and algorithms for causal analysis of stochastic and deterministic processes and for quantifying measurement uncertainties. It addresses methodological challenges in the analysis of time series and dynamical data, where correlation alone is insufficient to reveal the mechanisms governing system behavior. Manyapplications, therefore, require identifying causal relations between variables while reliably characterizing uncertainties arising from measurement processes, noise, and incomplete observations.The project will develop classical and modern approaches to causal analysis of time series based on probabilistic and statistical modeling, and integrate them with algorithms enabling statistical inference and prediction in the presence of randomness, measurement errors, and uncertainty.Modern applications in physical, biomedical, economic, environmental, and linguistic measurements, as well as in the social sciences (education, psychology), generate large and complex datasets with intricate dependence structures and temporal dynamics. A significant project component is hence the study of stochastic dynamical models, including diffusion processes, as a natural framework for modeling random dynamics observed via measurement time series. When modeling complex temporal or spatio-temporal data using kriging, causal structure will serve as a key starting point.The project also advances uncertainty methods for quantifying measurement uncertainties in line with modern metrology and aims to establish a unified methodological framework combining causal analysis, dynamical modeling, and statistical inference and forecasting. Interdisciplinary collaboration among the Institute of Measurement Science of the SAS, theMathematical Institute of the SAS, and the Faculty of Science of P. J. Šafárik University creates favorable conditions for the development of new theoretical results, efficient algorithms, and their applications.
Innovations in the Transfer Entropy Method: Implementing Alternative Entropic Measures for More Robust Causal Inference
Inovácie v metóde prenosovej entropie: Implementácia alternatívnych entropických mier pre robustnejšiu kauzálnu inferenciu
Program: Návratová projektová schéma
Duration: 1.7.2025 – 30.6.2026
Project leader: Mgr. Mezeiová Kristína, PhD.
Annotation: The aim of the project is to explore the use of alternative entropy measures, such as Rényi entropy, Tsallis entropy, and permutation entropy, in the transfer entropy method to enhance the accuracy, robustness, and computational efficiency of causal analysis in complex systems. The project will focus on the software implementation of appropriately modified causal algorithms, their testing on synthetic and real-world data, and the identification of areas where the proposed innovations provide significant advantages.