Thesis supervisor: András Hajdu
Location of studies (in Hungarian): University of Debrecen Faculty of Informatics Abbreviation of location of studies: DE IK
Description of the research topic:
Syllabus
The aim of the research is to design and optimize machine learning models for dedicated problems. The motivation lies in the fact that recently released and pre-trained efficient models (e.g. deep convolutional neural nets) are trained in general natural images, while for a specific task the important features can be more or less complex. That is, for optimal performance the necessary depth of the learning structures should be discovered first e.g. through starting the analysis of already available models. Based on these foundings a more efficient architecture can be proposed for a given application. The application fields based on primarily image information cover molecule design, Big Data analysis and clinical decision support.
Bibliography
1. Yoshua Bengio, Ian J. Goodfellow, Aaron Courville: Deep Learning, MIT Press, 2015.
2. Tariq Rashid: Make Your Own Neural Network (1st ed.). CreateSpace Independent Publishing Platform, USA, 2016.
3. Terrence J. Sejnowski: The Deep Learning Revolution, The MIT Press, 2018.
4. Suvrit Sra, Sebastian Nowozin, and Stephen J. Wright: Optimization for Machine Learning, The MIT Press, 2011.
Deadline for application: 2022-11-15
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
