Machine Learning Algorithms for Smart Systems

Principal Investigator: Márk Jelasity

A very important component of smart systems is the machine learning algorithms that fit models to collected data. Depending on the application area, designing algorithms may involve several challenges that include adapting to constraints of the computing infrastructure (centralized/decentralized, reliability, financial cost, etc), dealing with limitations of the data, such as limited access to labeled data records, and optimizing systems in terms of user-interaction via modeling user behavior. This research group will study machine learning problems related to such challenges in smart systems. In terms of computing infrastructure, we focus on decentralized platforms such as networks of mobile phones. We will carry out measurements over such platforms and design machine learning algorithms able to tolerate this harsh environment with data privacy in mind. In terms of lack of labeled data, we will study semi-supervised methods in the e-health domain. Finally, we will devise algorithms for analyzing user log data in several web environments.

Group's Content



Vanda Balog, Gábor Berend, Árpád Berta, Gábor Danner, József Dombi, Richárd Farkas, Abrar Hussain, Dániel Kövesi-Nagy, Gábor Kőrösi, István Megyeri, Tamás Németh, Gergely Pap, László Tóth

Close Menu