This graduate thesis explains and shows an example use of deep learning which is subpart of machine learning which is integral part of artificial intelligence. The difference between artificial intelligence, machine learning and deep learning is explained, each of those terms has been defined, also analysis of areas of application and what sort of problems they are used for, has been conducted.Machine learning was described in great detail, from history and development, to usage today and potential future. The division of machine learning to supervised, unsupervised, semi-supervised and reinforcement learning has been explained. Each type of learning uses different tehniques and are used for solving different problems which was analysed in this thesis. For every type of learning, most famous and widely used algorithms are listed and their structure, principle of work and usage was explained. Further development of machine learning which commonly use simple structures leads to deep learning which uses advanced models for solving complicated and real-world problems. Its tehniques have been analysed and architectures that are used to solve those kind of problems were explained.For experimental part, a software Faceswap was used, which uses the principle of deep learning with autoencoders to create deepfakes in which algorithm learns the features of presented face and then replaces it, in images or videos, with another desired face. After the explanation of how software works, it was used to swap faces in a video and visual proof was supplied to confirm the results of work.