Thesis supervisor: Miklós Hoffmann
Location of studies (in Hungarian): University of Debrecen Faculty of Informatics Abbreviation of location of studies: DEIK
Description of the research topic:
Machine Learning and its applications is an emerging field of informatics with numerous theories and algorithms which can be used for various purposes. The increase in computer power has enabled the widespread use of artificial intelligence and machine learning technologies, which are becoming part of our everyday lives, from image recognition, to automatic translation, from artificial intelligence assistants and chat-bots to autonomous cars, etc. Many of these applications are closely related to visualisation.
In many applications input data come from physical or virtual sensors. Processing large amount of data needs high performance computation, so the possible optimizations can have a high impact. Parallelization and ensemble methods can be used both for achieving higher performance and for reaching acceptable performance with machines having lower capacity. One of the well-known achievements of using ensemble learners is to get higher precision of prediction via combining of the predictions of the individual learners. It is also possible to use co-teaching and co-learning methods to use the positive effect of collaboration during the machine learning process, not only after it.
Sensor data can have noise which affects the results. Ensemble methods can be used to improve the input data quality by combining multiple measurements (Kalman-filter, averaging, error correction with neural network, etc.) Further crucial question is the visualisation of these learning methods and measurements in order to make the AI-based decision more transparent and understandable.
The aim of this PhD study is to make further contributions to these topics, specifically to improve ensemble-based methods in machine learning, to apply it in various fields, partly related to visualisation, and to support the understanding of AI through visualisation techniques.
Bibliography
1. Yoshua Bengio, Ian J. Goodfellow, Aaron Courville: Deep Learning, MIT Press, 2015.
2. Tibor Tajti and Benedek Nagy, "Motion sensor data correction using multiple sensors and multiple measurements," 2016 IEEE 14th International Symposium on Applied Machine Intelligence and Informatics (SAMI), 2016, pp. 287-291
3. Dietterich, Thomas G.: Ensemble learning. The handbook of brain theory and neural networks 2002, MIT Press, 110-125.
4. Guedj, Benjamin; Desikan, Bhargav Srinivasa. Pycobra: A python toolbox for ensemble learning and visualisation. Journal of Machine Learning Research, 2018, 18.190: 1-5.
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
