The article presents an analysis of modern methods of image generation: variational autoencoders (VAE), generative adversarial networks (GAN) and diffusion models. The main attention is paid to a comparative analysis of their performance, generation quality and computational requirements. The Frechet Inception Distance (FID) metric is used to assess the image quality. Diffusion models showed the best results (FID 20.8), outperforming VAE (FID 59.75) and GAN (FID 38.9), but require significant resources. VAEs are stable, but generate blurry images. GANs provide high quality, but suffer from training instability and mode collapse. Diffusion models, due to step-by-step noise decoding, combine detail and structure, which makes them the most promising. Also considered are methods of image-to-image generation used for image modification. The results of the study are useful for specialists in the field of machine learning and computer vision, contributing to the improvement of algorithms and expansion of the areas of application of generative models.

Keywords: deepfake, deep learning, artificial intelligence, GAN, VAE, diffusion model

Modern intelligent control systems (ICS) are complex software and hardware systems that use artificial intelligence, machine learning, and big data processing to automate decision-making processes. The article discusses the main tools and technologies used in the development of ICS, such as neural networks, deep learning algorithms, expert systems and decision support systems. Special attention is paid to the role of cloud computing, the Internet of Things and cyber-physical systems in improving the efficiency of intelligent control systems. The prospects for the development of this field are analyzed, as well as challenges related to data security and interpretability of models. Examples of the successful implementation of ICS in industry, medicine and urban management are given.

Keywords: intelligent control systems, artificial intelligence, machine learning, neural networks, big data, Internet of things, cyber-physical systems, deep learning, expert systems, automation

There are now many reasons why training in artificial intelligence (AI) technologies can be important for today's students. Therefore, AI-related disciplines are actively included by universities in undergraduate and graduate programs. It is important to teach students to understand how such technologies work and how they can be used to solve various problems. In turn, training is unthinkable without demonstrating examples of solving various problems. An important step in solving the problem of machine learning in general, incl. task of computer vision, is the stage of formation of the training sample. Therefore, the idea arose of writing a program that would be able to generate datasets on various topics for computer vision tasks. The data format of the generated sample for training train.csv is generally accepted and looks like this: each line is a description of one image; the first column contains the class labels to which the image belongs; the remaining columns contain the pixel values of the image, for example as a flat vector, where each value corresponds to the brightness of a corresponding pixel in the image. The resulting datasets can be used to organize the project activities of students on artificial intelligence.

Keywords: artificial intelligence, machine learning, computer vision, neural network, training set, dataset, C#, pixel, subpixel image processing, organization of student project activities