A new model of inventory control with returns is considered, when it is possible for consumers to return (under certain conditions) the products they have purchased and a similar model of multi-channel controlled queueing system with switching of service channels. An optimal inventory control strategy in such a system turns out to be four-level. Then we investigate a model of a multilinear queueing system (QS) with channel switching under uncertainty and study the analogy of this model with the model of inventory control with returns.

Object recognition is a branch of artificial vision and one of the pillars of machine vision. It consists in identifying the forms described in advance in a digital image and, in general, in a digital video stream. Although, as a rule, it is possible to perform recognition from video clips, the learning process is usually performed on images. In this paper, an algorithm for classifying and recognizing objects using convolutional neural networks is considered. The purpose of the work is to implement an algorithm for detecting and classifying various graphic objects fed from a webcam. The task is to first classify and recognize an object with high accuracy according to a given data set, and then demonstrate a way to generate images to increase the volume of the training data set by using a self-written generator. The classification and recognition algorithm used is invariant to transfer, shift and rotation. A significant novelty of this work is the creation of a self-written generator that allows using various types of augmentation (artificial increase in the volume of the training sample by modifying the training data) to form new groups of modified images each time.

Steerable concentric continuum robots are robots with a flexible structure that are able to bend at any point. Such robots consist of tubes inserted inside each other. The shape and length of the robot can be varied by controlling relative translations, rotations and bending angles of the tubes. This feature allows them to operate in confined working areas such as human heart, lungs, nuclear reactors and so on. However, existing solutions to the inverse kinematics for these robots have the following problems: high computational cost, singularity problems, complex matrix calculations, inability to control a robot tip orientation or requirement to go through the configuration parameters. This paper presents a modification of the Forward And Backward Reaching Inverse Kinematics (FABRIK) algorithm for multi-section continuum robots. Particularly, we applied the proposed modification of the FABRIK algorithm to the steerable concentric continuum robots.

Objective. To assess the characteristics of the conducting pathways in the white matter in patients with types 1 and 2 diabetes mellitus (DM) with and without cognitive impairment. Materials and methods. The study involved 85 patients with type 1 diabetes mellitus and 135 with type 2 diabetes mellitus, divided into patients with normal cognitive functions and those with cognitive impairments. The groups were comparable in terms of age and disease duration. Screening for cognitive disorders was with the Montreal Cognitive Assessment (MocA). Brain MRI scans were performed using a 1.5-T instrument. Results. These studies identified a predominance of mild and moderate cognitive impairments in type 1 DM and moderate and severe in type 2 DM; impairments were apparent mainly as impairments to memory, attention, and visuospatial orientation. Between-group analysis of brain tractography studies did not identify any differences in the integrity of tracts in types 1 and 2 DM, though very significant risk factors for impaired functioning of the conducting pathways were seen: arterial hypertension (H = 6.602833, p = 0.0368), the severity of polyneuropathy (H = 15.30420, p = 0.0005), the extent of nephropathy (H = 9.993923, p = 0.0068), and the extent of retinopathy (H = 8.445891, p = 0.0376) in type 1 DM; age (H = 7.381742, p = 0.0607) in type 2 DM; the cholesterol level (H = 4.009380, p = 0.0452; H = 4.057357, p = 0.0440) in both types of DM. The corticospinal and commissural tracts were the most susceptible to damage. Conclusions. No statistically significant differences in the axial diffusion of brain tracts were seen in patients with and without cognitive impairments. However, we confirmed the most important risk factors for damage to white matter structures: arterial hypertension, complications of diabetes, cholesterol level, and age.

The chapter deals with the construction of generative models using Variational Autoencoder (VAE) and Generative Adversarial Neural Networks to synthesize new medical data. VAE is a synthesis of two complete neural networks: an encoder E and a generator G, as well as the latent space connecting them and enabling them to carry out random transformation and interpolation. Generative Adversarial Nets (GAN) in their turn are built on the principle of interaction between a generative model (generator G) and a discriminating model (discriminator D). When creating generator G (both VAE and GAN), its architecture of a neural network based on convolutional layers, with the application of the new deep learning framework Tensorflow-addons is used. As E and D encoders, respectively, the models of transfer learning, problem domain-image feature vector are used in the work. The comparison between them is made in the chapter and the most optimal model for solving the proposed problem is selected. The chapter presents the results of the research obtained on the basis of VAE and GAN implementation.