Thesis supervisor: Attila Fazekas
Location of studies (in Hungarian): University of Debrecen, Faculty of Informatics Abbreviation of location of studies: ITDI
Description of the research topic:
Binary classification and multi-class classification are fraught with several issues. Firstly, the datasets used for training are often unbalanced, noisy, and/or not of ideal size. Secondly, there have been several problems identified in determining the performance scores that express the efficiency of classifiers in recent times.
Based on the above, it is worth investigating the behavior, noise sensitivity, and database dependency of classifiers. Furthermore, the research area that deals with the consistency of performance scores used to characterize classifiers is also fraught with numerous challenges. The doctoral topic primarily addresses these areas.
Bibliography
1. Marcel Katulumba Mbiya: Comparative study of set methods for classification: Application of Adaboosting and Random Forest to Binary and Multi-class databases, Our Knowledge Publishing, 2022.
2. Francisco Herrera, Francisco Charte, Antonio J. Rivera, María J. del Jesus: Multilabel Classification, Springer, 2016.
3. Szeghalmy, S.; Fazekas, A. A Highly Adaptive Oversampling Approach to Address the Issue of Data Imbalance. Computers 2022, 11, 73.
4. G. Kovács, A. Fazekas, A new baseline for retinal vessel segmentation: Numerical identification and correction of methodological inconsistencies affecting 100+ papers, Medical Image Analysis 75 (2022) 102300.
Deadline for application: 2024-05-15
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
