Thesis supervisor: András Lőrincz
Location of studies (in Hungarian): ELTE Faculty of Informatics Abbreviation of location of studies: ELTE
Description of the research topic:
Deep neural networks brought about a breakthrough in the field of artificial intelligence. Their superhuman performance, however, is restricted to a few subfields, their generalization capability is limited and they need behavioral supervision. Promising solution emerges by using a number of deep network in such a way that (i) all of them learn to predict, (ii) they also learn from the prediction errors, and (iii) the networks are able to correct the errors by means of Bayes principle. Prediction means model learning, the error of the model is the anomaly and thus the constraint of learning and, in turn, the basis of training each other. A number of databases, mostly videos are available for the integration of different dataflows (in case of speech and video an example is lip reading) for the training of the joint model.
Required language skills: fluent English Further requirements: Excellent knowledge in mathematics, including analysis and linear algebra
High level programming skills in C++ and Python
Hard working in a competitive field
Precision and flexibility for teamwork
Number of students who can be accepted: 1
Deadline for application: 2018-05-31
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
