Evaluation of a snip pruning method for a state-of-the-art face detection model

With rapid development of machine learning and subsequently deep learning, deep neural networks achieved remarkable results in solving various tasks. However, with increasing the accuracy of trained models, new architectures of neural networks present new challenges as they require significant amount of computing power for training and inference. This paper aims to review existing approaches to reducing computational power and training time of the neural network, evaluate and improve one of existing pruning methods for a face detection model.

Solving the Cauchy problem for an elliptic equation using Bat Algorithm

This paper presents a method for solving a class of inverse problems for elliptic equations known as the data completion problem.  The goal is to recover missing data on the inaccessible part of the boundary using measurements from the accessible part.  The inherent difficulty of this problem arises from its ill-posed nature, as it is susceptible to variations in the input data.  To address this challenge, the proposed approach integrates Tikhonov regularization to enhance the stability of the problem.  To solve this problem, we use a metaheuristic approach, specificall


The architecture of the unified system of centralized control and measurements of the SKS-7 network is outdated, the technologies used to organize its work and implement data storage need to be updated and optimized, because the current state of the system causes a decrease in speed and performance, as well as a critical drop in reliability. Considering the factors, there is also a need to modernize the software and hardware of the IP protocol analysis and monitoring system.

Modeling of the modes of operation of wind energy installations in hybrid power supply systems

The article presents modern schemes for the organization of wind-solar power supply systems. Available approaches to managing the energy-dynamic process of operation of wind power plants as part of hybrid power supply systems are given, and modern research on this topic is given. The results of the development of a mathematical model of the energy-dynamic processes of the hybrid wind-solar power supply system, which includes wind power plants, solar panels, and a battery energy storage system. The universal structural diagram of such a system is substantiated.

Genetic algorithm parenting fitness

The evolution scheme phase, in which the genetic algorithms select individuals that will form the new population, had an important impact on these algorithms.  Many approaches exist in the literature.  However, these approaches consider only the value of the fitness function to differenciate best solutions from the worst ones.  This article introduces the parenting fitness, a novel parameter, that defines the capacity of an individual to produce fittest offsprings.  Combining the standard fitness function and the parenting fitness helps the genetic algorithm to be more efficient, hence, pro

Optimization of the infrastructure of the distributed information system of goods accounting

An existing goods accounting information system was assessed for possible infrastructure optimization. A various parts of the system were analyzed to improve infrastructure costs without having a significant degradation of non-functional requirements. Modeling of the optimized system was performed, and evaluation of the infrastructure costs was made. Several optimization directions were evaluated, analyzed and either recommended or rejected.

Optimization of Thermal Subsystem of Thermo Transducers for Measuring the Temperature of Gas Flows

The error of measuring the temperature of gas flows by contact methods consists of two main components: 1) the error that occurs in the process of measuring the physical quantity into which the temperature is converted (the error of the measuring means); 2) errors arising in the process of converting the temperature into a measured physical value using a thermo transducer.

Optimization of Hydrolysis in Ethanol Production from Bamboo

This research involved optimizing acid hydrolysis in the development of ethanol, a promising alternative energy source for restricted crude oil, from lignocellulosic materials (bamboo). The conversion of bamboo to ethanol can mainly be accomplished through three process steps: pretreatment of bamboo wood for the removal of lignin and hemicellulose, acid hydrolysis of pretreated bamboo for the conversion of cellulose into sugar reduction (glucose) and fermentation of sugars into ethanol using anaerobic Saccharomyces cerevisiae.

Optimization of Isocratic Ion Chromatography Methods for Simultaneous Inorganic Anions and Carboxylic Acids Determination

As a part of preliminary studies, the possibility of quantitative oxidation of aldehydes to carboxylic acids and their determination by isocratic ion chromatography with conductometric detection has been tested.

Density based fuzzy support vector machine: application to diabetes dataset

In this work, we propose a deep prediction diabetes system based on a new version of the support vector machine optimization model.  First, we determine three types of patients (noisy, cord, and interior) basing on specific parameters. Second, we equilibrate the clinical data sets by suppressing noisy and cord patients.  Third, we determine the support vectors by solving an optimization program with a reasonable size.