Thesis supervisor: Ferenc Kruzslicz
co-supervisor: Miklós Hornyák
Location of studies (in Hungarian): H-7622 Pécs, Rákóczi út 80. Abbreviation of location of studies: KTK
Description of the research topic:
Transfer learning is a machine learning approach to adapt previously acquired knowledge to new problems where the development of native models are impossible or too expensive. In a constantly changing economic environment automatic decision making model must be not only continuously monitored about their performance, but also updated and adapted the changes. Transfer learning is most useful when there an trained model exists, optimized for a larger population (region) which must be justified for a smaller population having not enough data to develop an independent model for itself. Transfer learning originates from text mining, but because it is a general-purpose approach, it also enables creating novel economic models.
Possible research directions:
• There are many open questions regarding transfer learning: how to measure its effectiveness, how to prevent the phenomenon of negative or even overfitting, or how much environmental changes it can follow.
• By expanding the range of economic applications of this methodology, doctoral students can create special models, develop transfer learning strategies or even develop transfer learning algorithms.
Number of students who can be accepted: 1
Deadline for application: 2024-09-30
2024. IV. 17. ODT ülés Az ODT következő ülésére 2024. június 14-én, pénteken 10.00 órakor kerül sor a Semmelweis Egyetem Szenátusi termében (Bp. Üllői út 26. I. emelet).
Login
