Search

Handwritten signatures are used in authentication systems as a universal biometric identifier. Signature authenticity verification requires building and training a classifier. This paper describes a new approach to the verification of handwritten signatures by dynamic characteristics with a fuzzy rule-based classifier. It is suggested to use the metaheuristic Gravitational Search Algorithm for the selection of the relevant features and tuning fuzzy rule parameters. The efficiency of the approach was tested with an original dataset; the type II errors in finding the signature authenticity did not exceed 0.5% for the worst model and 0.08% for the best model.

A promising approach to overcome the various shortcomings of password systems is the use of biometric authentication, in particular the use of electroencephalogram (EEG) data. In this paper, we propose a subject-independent learning method for EEG-based biometrics using Hilbert spectrograms of the data. The proposed neural network architecture treats the spectrogram as a collection of one-dimensional series and applies one-dimensional dilated convolutions over them, and a multi-similarity loss was used as the loss function for subject-independent learning. The architecture was tested on the publicly available PhysioNet EEG Motor Movement/Imagery Dataset (PEEGMIMDB) with a 14.63% Equal Error Rate (EER) achieved. The proposed approach’s main advantages are subject independence and suitability for interpretation via created spectrograms and the integrated gradients method.

This paper compares neural networks, specifically Unet, MobileNetV2, VGG16 and YOLOv4-tiny, for image segmentation as part of a study aimed at finding an optimal solution for price tag data analysis. The neural networks considered were trained on an individual dataset collected by the authors. Additionally, this paper covers the automatic image text recognition approach using EasyOCR API. Research revealed that the optimal network for segmentation is YOLOv4-tiny, featuring a cross validation accuracy of 96.92%. EasyOCR accuracy was also calculated and is 95.22%.

Information security is one of the most important attributes of distributed systems that often operate on unreliable networks. Enabling security features during the development of a distributed system requires the careful analysis of potential attacks or threats in different contexts, a process often referred to as «threat modeling». Information protection should be comprehensive, but it is also necessary to take into account the possibility of the emergence of threats specific to a certain information system. Many public and private organizations are still trying to implement system models and the threats directed at them on their own. The main reason for this is the lack of useful and high-quality methodologies that can help developers design system models. This review explores a variety of the literature on confidentiality- and integrity-aware system design methodologies, as well as threat classification methods, and identifies key issues that may be referenced by organizations to make design system processes easier. In particular, this article takes a look at the extent to which existing methodologies cover objects of protection and methods of classifying threats, as well as whether there are such models of systems in which the object itself and the threats directed at it are described. This includes whether the compiled models exhibit symmetry or asymmetry. This literature research shows that methodologies appear to be heterogeneous and versatile, since existing methodologies often only focus on one object of protection (a system). Based on the given analysis, it can be concluded that the existing methodologies only relate superficially to the description of system models and threats, and it is necessary to develop a more complete abstract model of the protected object and threats aimed at it in order to make this model suitable for any organization and protect it against most threats.

Machine learning algorithms based on neural networks are vulnerable to adversarial attacks. The use of attacks against authentication systems greatly reduces the accuracy of such a system, despite the complexity of generating a competitive example. As part of this study, a white-box adversarial attack on an authentication system was carried out. The basis of the authentication system is a neural network perceptron, trained on a dataset of frequency signatures of sign. For an attack on an atypical dataset, the following results were obtained: with an attack intensity of 25%, the authentication system availability decreases to 50% for a particular user, and with a further increase in the attack intensity, the accuracy decreases to 5%.

Almost all industrial internet of things (IIoT) attacks happen at the data transmission layer according to a majority of the sources. In IIoT, different machine learning (ML) and deep learning (DL) techniques are used for building the intrusion detection system (IDS) and models to detect the attacks in any layer of its architecture. In this regard, minimizing the attacks could be the major objective of cybersecurity, while knowing that they cannot be fully avoided. The number of people resisting the attacks and protection system is less than those who prepare the attacks. Well-reasoned and learning-backed problems must be addressed by the cyber machine, using appropriate methods alongside quality datasets. The purpose of this paper is to describe the development of the cybersecurity datasets used to train the algorithms which are used for building IDS detection models, as well as analyzing and summarizing the different and famous internet of things (IoT) attacks. This is carried out by assessing the outlines of various studies presented in the literature and the many problems with IoT threat detection. Hybrid frameworks have shown good performance and high detection rates compared to standalone machine learning methods in a few experiments. It is the researchers’ recommendation to employ hybrid frameworks to identify IoT attacks for the foreseeable future.

The determining factor in the accelerated pace of informatization is the increase in the speed and reliability of data transmission networks. In this regard, new and existing standards are developed and modernized. A lot of organizations are constantly working on the development and implementation of new generation communication networks. This article provides an overview of available software solutions that allow us to investigate and evaluate the behavior of data networks. In particular, tools suitable for mobile communication systems were determined, having sufficient built-in functionality and allowing us to add our own implementations. NS3 has been chosen as a suitable network simulator. Apart from the review, a solution for this tool was developed. It allows estimating the reliability of data transmission from the start movement of a network node at all times during its removal from a base station.

In recent years, a lot of IoT devices, wireless sensors, and smart things contain information that must be transmitted to the server for further processing. Due to the distance between devices, battery power, and the possibility of sudden device failure, the network that connects the devices must be scalable, energy efficient, and flexible. Particular attention must be paid to the protection of the transmitted data. The Bluetooth mesh was chosen as such a network. This network is built on top of Bluetooth Low-Energy devices, which are widespread in the market and whose radio modules are available from several manufacturers. This paper presents an overview of security mechanisms for the Bluetooth mesh network. This network provides encryption at two layers: network and upper transport layers, which increases the level of data security. The network uses sequence numbers for each message to protect against replay attacks. The introduction of devices into the network is provided with an encryption key, and the out-of-band (OOB) mechanism is also supported. At the moment, a comparison has been made between attacks and defense mechanisms that overlap these attacks. The article also suggested ways to improve network resiliency.

The need to assess the risks of the trustworthiness of counterparties is increasing every year. The identification of increasing cases of unfair behavior among counterparties only confirms the relevance of this topic. The existing work in the field of information and economic security does not create a reasonable methodology that allows for a comprehensive study and an adequate assessment of a counterparty (for example, a developer company) in the field of software design and development. The purpose of this work is to assess the risks of a counterparty’s trustworthiness in the context of the digital transformation of the economy, which in turn will reduce the risk of offenses and crimes that constitute threats to the security of organizations. This article discusses the main methods used in the construction of a mathematical model for assessing the trustworthiness of a counterparty. The main difficulties in assessing the accuracy and completeness of the model are identified. The use of cross-validation to eliminate difficulties in building a model is described. The developed model, using machine learning methods, gives an accurate result with a small number of compared counterparties, which corresponds to the order of checking a counterparty in a real system. The results of calculations in this model show the possibility of using machine learning methods in assessing the risks of counterparty trustworthiness.

Modern facial recognition algorithms make it possible to identify system users by their appearance with a high level of accuracy. In such cases, an image of the user’s face is converted to parameters that later are used in a recognition process. On the other hand, the obtained parameters can be used as data for pseudo-random number generators. However, the closeness of the sequence generated by such a generator to a truly random one is questionable. This paper proposes a system which is able to authenticate users by their face, and generate pseudo-random values based on the facial image that will later serve to generate an encryption key. The generator of a random value was tested with the NIST Statistical Test Suite. The subsystem of image recognition was also tested under various conditions of taking the image. The test results of the random value generator show a satisfactory level of randomness, i.e., an average of 0.47 random generation (NIST test), with 95% accuracy of the system as a whole.

As part of the study, the problem of developing and applying a metric for assessing the degree of similarity of time series is considered, which makes it possible to consider the assumptions about the model of a series when comparing, as well as to compare the values of the corresponding characteristics of the series. Characteristics can be values that describe the structure of a series, or directly the values of the approximating function, which can be obtained using nonparametric statistics methods. One of the directions in which this approach can be applied to assessing the similarity of time series is the study of vocal performances. In practice, the degree of similarity in the performance of melodies by several speakers was analyzed. It was determined that, using the synchronicity metric, it is possible to implement an exercise in which students need to repeat the melody after the teacher. In addition, this approach was applied in the segment identification module with an abrupt change in the sounding of the fundamental frequency. This work is devoted to the modification of the program complex for vocal recognition in order to identify notes with a sharp change in the fundamental frequency. The complex is aimed at carrying out additional independent training in teaching vocals. The use of the software package will allow, in real time, providing feedback to the user with an assessment of the quality of their singing. This should allow students to study not only under the supervision of a teacher, but also independently in the early stages of learning. The basic algorithm of the program recognizes notes without sharp changes in frequencies with high accuracy, which is confirmed by experiments. In order to recognize by the algorithms of the program notes sung vibrato and glissando in singing, a new analysis method based on the metric of time series synchronicity is proposed.

In Tomsk University of Control Systems and Radioelectronics (TUSUR) one of the main areas of research is information security. The work is carried out by a scientific group under the guidance of Professor Shelupanov. One of the directions is the development of a comprehensive approach to assessing the security of the information systems. This direction includes the construction of an information security threats model and a protection system model, which allow to compile a complete list of threats and methods of protection against them. The main directions of information security tools development are dynamic methods of biometrics, methods for generating prime numbers for data encryption, steganography, methods and means of data protection in Internet of Things (IoT) systems. The article presents the main results of research in the listed areas of information security. The resultant properties in symmetric cryptography are based on the properties of the power of the generating functions. The authors have obtained symmetric principles for the development of primality testing algorithms, as discussed in the Appendix.

This paper concerns several important topics of the Symmetry journal, namely, pattern recognition, computer-aided design, diversity and similarity. We also take advantage of the symmetric and asymmetric structure of a transfer function, which is responsible to map a continuous search space to a binary search space. A new method for design of a fuzzy-rule-based classifier using metaheuristics called Gravitational Search Algorithm (GSA) is discussed. The paper identifies three basic stages of the classifier construction: feature selection, creating of a fuzzy rule base and optimization of the antecedent parameters of rules. At the first stage, several feature subsets are obtained by using the wrapper scheme on the basis of the binary GSA. Creating fuzzy rules is a serious challenge in designing the fuzzy-rule-based classifier in the presence of high-dimensional data. The classifier structure is formed by the rule base generation algorithm by using minimum and maximum feature values. The optimal fuzzy-rule-based parameters are extracted from the training data using the continuous GSA. The classifier performance is tested on real-world KEEL (Knowledge Extraction based on Evolutionary Learning) datasets. The results demonstrate that highly accurate classifiers could be constructed with relatively few fuzzy rules and features.