Our research revolves around four main areas: network science, educational data science, health data analytics, and intelligent monitoring & anomaly detection.
Network science
Our network science research covers both fundamental methodological research and applied research. Our main research areas include the structural characterization of complex networks, particularly fractality and robustness, data-driven analysis of fractal networks, calibrating network models, graph embedding methods, social network analysis, co-authorship network analysis.
Educational data science is an important branch of data science that aims to extract knowledge from various forms of massive educational data using statistical and machine learning methods. In cooperation with the Central Academic Office of BME, we aim to assist educational stakeholders by providing a better understanding of the big data stored in the educational administrative system. Our expertise includes identifying students at risk of dropping out, assessing the predictive validity of the admission system, identifying various factors of student success, quantifying the impact of interventions, explainable artificial intelligence.
We are committed to using the tools of network science and data science in medicine. We develop decision-support tools that assist the physicians by predictions based on data available at hospitalization.
Intelligent monitoring, state prediction, and anomaly detection
An important line of our research revolves around high-dimensional, high-frequency data. We do research about intelligent monitoring and predictive maintenance, particularly anomaly detection. We developed a coupla-based anomaly detection and localization algorithm that performs well on high-dimensional data and can cope with missing values as well.