Швай, НадіяКрошин, Олександр2022-01-172022-01-172021https://ekmair.ukma.edu.ua/handle/123456789/22284Rapid developments in the Deep Learning domain in recent years let researchers and practitioners shift their focus from training machine learning models itself to transferring the already-learnt knowledge and applying it in different applications. This paper discusses Domain Adaptation, a subdomain of transfer learning, primarily aimed at applying knowledge from a given source domain to an unknow target one. It discusses various Domain Adaptation settings under the context of Computer Vision, introduces self-ensembling Domain Adaptation methods for semi-supervised learning and illustrates its capabilities with proper experiments. Experiments were implemented with Python 3.6 using libraries pytorch, numpy, pandas, opencv, matplotlib, torch-salad, etc.enадаптаціякласифікація зображеньмагістерська роботаOther