{"id":11005,"date":"2021-10-18T11:35:06","date_gmt":"2021-10-18T09:35:06","guid":{"rendered":"https:\/\/kinit.sk\/member\/peter-pavlik\/"},"modified":"2025-10-19T16:43:18","modified_gmt":"2025-10-19T14:43:18","slug":"peter-pavlik","status":"publish","type":"member","link":"https:\/\/kinit.sk\/sk\/clen\/peter-pavlik\/","title":{"rendered":"Peter Pavl\u00edk"},"content":{"rendered":"
\n

Peter is a research assistant focused on machine learning with experience mainly in time series prediction using both statistical and artificial intelligence approaches, ensemble learning and using convolutional neural networks with emphasis on interpretability. He worked on multiple projects in the energy domain and is interested in developing environmentally friendly solutions of the future.<\/p>\n<\/div>\n\n

\n

Peter holds a Masters\u2019 degree in Intelligent Software Systems from the Faculty of Informatics and Information Technologies STU. During his bachelor studies at FIIT, he was a member of the research oriented group of students, received the Best Bachelor Paper Award at the IIT.SRC 2018 faculty student research conference and graduated cum laude. After graduating, he worked as a data engineer in an international company where he gained industry knowledge on development and maintenance of Big Data processes.<\/p>\n<\/div>\n\n

\n

PhD topic: Physics-informed Deep Learning in Forecasting<\/h3>\n<\/div>\n\n
\n

Supervising team:<\/strong> Viera Rozinajov\u00e1<\/a> (KInIT), Anna Bou Ezzeddine<\/a> (KInIT), Softec<\/a> (industry partner)<\/p>\n<\/div>\n\n

\n

We explore the use of deep learning to improve the efficiency and accuracy of forecasts of physical systems. Accurate forecasting of changes occurring in the real world is inseparable from modeling and simulation of underlying physical processes. This was traditionally performed using complex numerical physical simulations. Data-driven deep learning emulators are a promising alternative to these numerical approaches, showing many advantages in some areas but also lacking in others. Machine learning methods incorporating existing domain knowledge about the modeled physical processes can combine the best aspects of both paradigms. This is the premise of physics-informed machine learning (PIML). We apply it in the weather forecasting domain, working closely with the company SOFTEC and the Slovak Hydrometeorological Institute. We train models for the task of precipitation nowcasting from radar reflectivity data, with a focus on correctly estimating the precipitation quantity in the solid phase for the purposes of optimizing winter road maintenance.<\/p>\n<\/div>\n\n

\n
\n