All issues

The reconstruction of charged particle trajectories in tracking detectors is a key problem in the analysis of experimental data for high energy and nuclear physics.

The amount of data in modern experiments is so large that classical tracking methods such as Kalman filter can not process them fast enough. To solve this problem, we have developed two neural network algorithms of track recognition, based on deep learning architectures, for local (track by track) and global (all tracks in an event) tracking in the GEM tracker of the BM@N experiment at JINR (Dubna). The advantage of deep neural networks is the ability to detect hidden nonlinear dependencies in data and the capability of parallel execution of underlying linear algebra operations.

In this work we generalize these algorithms to the cylindrical GEM inner tracker of BESIII experiment. The neural network model RDGraphNet for global track finding, based on the reverse directed graph, has been successfully adapted. After training on Monte Carlo data, testing showed encouraging results: recall of 98% and precision of 86% for track finding.

The local neural network model TrackNETv2 was also adapted to BESIII CGEM successfully. Since the tracker has only three detecting layers, an additional neuro-classifier to filter out false tracks have been introduced. Preliminary tests demonstrated the recall value at the first stage of 99%. After applying the neuro-classifier, the precision was 77% with a slight decrease of the recall to 94%. This result can be improved after the further model optimization.

We suggested some approaches for solving image processing tasks in the decision support system (DSS) of the Center for Automated Recording of Administrative Offenses of the Road Traffic (CARAO). The main task of this system is to assist the operator in obtaining accurate information about the vehicle registration plate and the vehicle brand/model based on images obtained from the photo and video recording systems. We suggested the approach for vehicle registration plate recognition and brand/model classification on the images based on modern neural network models. LPRNet neural network model supplemented by Spatial Transformer Layer was used to recognize the vehicle registration plate. The ResNeXt-101-32x8d neural network model was used to classify for vehicle brand/model. We suggested the approach to construct the training set for the neural network of vehicle registration plate recognition. The approach is based on computer vision methods and machine learning algorithms. The SIFT algorithm was used to detect and describe local features on images with the vehicle registration plate. DBSCAN clustering was used to detect and delete outliers in such local features. The accuracy of vehicle registration plate recognition was 96% on the testing set. We suggested the approach to improve the efficiency of using the ResNeXt-101-32x8d model at additional training and classification stages. The approach is based on the new architecture of convolutional neural networks with “freezing” weight coefficients of convolutional layers, an additional convolutional layer for parallelizing the classification process, and a set of binary classifiers at the output. This approach significantly reduced the time of additional training of neural network when new vehicle brand/model classification was needed. The final accuracy of vehicle brand/model classification was 99% on the testing set. The proposed approaches were tested and implemented in the DSS of the CARAO of the Republic of Tatarstan.

Cognitive and regulatory processes in organism are ensured by the functioning of several different network systems — neural, endocrine, immune, and gene ones. These systems are, however, closely related and form a single integrated neurogenohumoral cognitive-regulatory dynamic system of organism. A review of publications is given which shows that it is possible to associate with this dynamic system a corresponding Lyapunov function (energy function, potential function) and that analyzing this function allows, due to its geometrical insight, to easily discover a set of general properties of cognitive and regulatory functioning of organism.

Automatic recognition of personal identity by biometric features is based on unique peculiarities or characteristics of people. Biometric identification process consist in making of reference templates and comparison with new input data. Iris pattern recognition algorithms presents high accuracy and low identification errors percent on practice. Iris pattern advantages over other biometric features are determined by its high degree of freedom (nearly 249), excessive density of unique features and constancy. High recognition reliability level is very important because it provides search in big databases. Unlike one-to-one check mode that is applicable only to small calculation count it allows to work in one-to-many identification mode. Every biometric identification system appears to be probabilistic and qualitative characteristics description utilizes such parameters as: recognition accuracy, false acceptance rate and false rejection rate. These characteristics allows to compare identity recognition methods and asses the system performance under any circumstances. This article explains the mathematical model of iris pattern biometric identification and its characteristics. Besides, there are analyzed results of comparison of model and real recognition process. To make such analysis there was carried out the review of existing iris pattern recognition methods based on different unique features vector. The Python-based software package is described below. It builds-up probabilistic distributions and generates large test data sets. Such data sets can be also used to educate the identification decision making neural network. Furthermore, synergy algorithm of several iris pattern identification methods was suggested to increase qualitative characteristics of system in comparison with the use of each method separately.

The proteins of the Major Histocompatibility Complex (MHC) play a key role in the functioning of the adaptive immune system, and the identification of peptides that bind to them is an important step in the development of vaccines and understanding the mechanisms of autoimmune diseases. Today, there are a number of methods for predicting the binding of a particular MHC allele to a peptide. One of the best such methods is NetMHCpan-4.0, which is based on an ensemble of artificial neural networks. This paper presents a methodology for qualitatively improving the underlying neural network underlying NetMHCpan-4.0. The proposed method uses the ensemble construction technique and adds as input an estimate of the Potts model taken from static mechanics, which is a generalization of the Ising model. In the general case, the model reflects the interaction of spins in the crystal lattice. Within the framework of the proposed method, the model is used to better represent the physical nature of the interaction of proteins included in the complex. To assess the interaction of the MHC + peptide complex, we use a two-dimensional Potts model with 20 states (corresponding to basic amino acids). Solving the inverse problem using data on experimentally confirmed interacting pairs, we obtain the values of the parameters of the Potts model, which we then use to evaluate a new pair of MHC + peptide, and supplement this value with the input data of the neural network. This approach, combined with the ensemble construction technique, allows for improved prediction accuracy, in terms of the positive predictive value (PPV) metric, compared to the baseline model.

This article solves the problem of developing a technology for collecting initial data for building models for assessing the functional state of a person. This condition is assessed by the pupil response of a person to a change in illumination based on the pupillometry method. This method involves the collection and analysis of initial data (pupillograms), presented in the form of time series characterizing the dynamics of changes in the human pupils to a light impulse effect. The drawbacks of the traditional approach to the collection of initial data using the methods of computer vision and smoothing of time series are analyzed. Attention is focused on the importance of the quality of the initial data for the construction of adequate mathematical models. The need for manual marking of the iris and pupil circles is updated to improve the accuracy and quality of the initial data. The stages of the proposed technology for collecting initial data are described. An example of the obtained pupillogram is given, which has a smooth shape and does not contain outliers, noise, anomalies and missing values. Based on the presented technology, a software and hardware complex has been developed, which is a collection of special software with two main modules, and hardware implemented on the basis of a Raspberry Pi 4 Model B microcomputer, with peripheral equipment that implements the specified functionality. To evaluate the effectiveness of the developed technology, models of a single-layer perspetron and a collective of neural networks are used, for the construction of which the initial data on the functional state of intoxication of a person were used. The studies have shown that the use of manual marking of the initial data (in comparison with automatic methods of computer vision) leads to a decrease in the number of errors of the 1st and 2nd years of the kind and, accordingly, to an increase in the accuracy of assessing the functional state of a person. Thus, the presented technology for collecting initial data can be effectively used to build adequate models for assessing the functional state of a person by pupillary response to changes in illumination. The use of such models is relevant in solving individual problems of ensuring transport security, in particular, monitoring the functional state of drivers.

The numerical methods used to simulate the evolution of hydrodynamic systems require the considerable use of computational resources thus limiting the number of possible simulations. The data-driven simulation technique is one promising approach to the development of heuristic models, which may speed up the study of such models. In this approach, machine learning methods are used to tune the weights of an artificial neural network that predicts the state of a physical system at a given point in time based on initial conditions. This article describes an original neural network architecture and a novel multi-stage training procedure which create a heuristic model of a two-phase flow in a heterogeneous porous medium. The neural network-based model predicts the states of the grid cells at an arbitrary timestep (within the known constraints), taking in only the initial conditions: the properties of the heterogeneous permeability of the medium and the location of sources and sinks. The proposed model requires orders of magnitude less processor time in comparison with the classical numerical method, which served as a criterion for evaluating the effectiveness of the trained model. The proposed architecture includes a number of subnets trained in various combinations on several datasets. The techniques of adversarial training and weight transfer are utilized.

When modeling turbulent flows in practical applications, it is often necessary to carry out a series of calculations of bodies of similar topology. For example, bodies that differ in the shape of the fairing. The use of convolutional neural networks allows to reduce the number of calculations in a series, restoring some of them based on calculations already performed. The paper proposes a method that allows to apply a convolutional neural network regardless of the method of constructing a computational mesh. To do this, the flow field is reinterpolated to a uniform mesh along with the body itself. The geometry of the body is set using the signed distance function and masking. The restoration of the flow field based on part of the calculations for similar geometries is carried out using a neural network of the UNet type with a spatial attention mechanism. The resolution of the nearwall region, which is a critical condition for turbulent modeling, is based on the equations obtained in the nearwall domain decomposition method.

A demonstration of the method is given for the case of a flow around a rounded plate by a turbulent air flow with different rounding at fixed parameters of the incoming flow with the Reynolds number $Re = 10^5$ and the Mach number $M = 0.15$. Since flows with such parameters of the incoming flow can be considered incompressible, only the velocity components are studied directly. The flow fields, velocity and friction profiles obtained by the surrogate model and numerically are compared. The analysis is carried out both on the plate and on the rounding. The simulation results confirm the prospects of the proposed approach. In particular, it was shown that even if the model is used at the maximum permissible limits of its applicability, friction can be obtained with an accuracy of up to 90%. The work also analyzes the constructed architecture of the neural network. The obtained surrogate model is compared with alternative models based on a variational autoencoder or the principal component analysis using radial basis functions. Based on this comparison, the advantages of the proposed method are demonstrated.

Formation of adequate behavioral patterns in condition of the unknown environment carried out through exploratory behavior. At the same time the rapid formation of an acceptable pattern is more preferable than a long elaboration perfect pattern through repeat play learning situation. In extreme situations, phenomenon of imprinting is observed — instant imprinting of behavior pattern, which ensure the survival of individuals. In this paper we propose a hypothesis and imprint model when trained on a single successful pattern of virtual robot's neural network demonstrates the effective functioning. Realism of the model is estimated by checking the stability of playback behavior pattern to perturbations situation imprint run.

Working memory mechanisms in spiking neural networks consisting of leaky integrate-and-fire neurons with adaptive threshold and synaptic plasticity are studied in this work. Moderate size networks including thousands of neurons were explored. Working memory is a network ability to keep in its state the information about recent stimuli presented to the network such that this information is sufficient to determine which stimulus has been presented. In this study, network state is defined as the current characteristics of network activity only — without internal state of its neurons. In order to discover the neuronal structures serving as a possible substrate of the memory mechanism, optimization of the network parameters and structure using genetic algorithm was carried out. Two kinds of neuronal structures with the desired properties were found. These are neuron pairs mutually connected by strong synaptic links and long tree-like neuronal ensembles. It was shown that only the neuron pairs are suitable for efficient and reliable implementation of working memory. Properties of such memory units and structures formed by them are explored in the present study. It is shown that characteristics of the studied two-neuron memory units can be set easily by the respective choice of the parameters of its neurons and synaptic connections. Besides that, this work demonstrates that ensembles of these structures can provide the network with capability of unsupervised learning to recognize patterns in the input signal.