All issues

In today’s technology-driven world, lite devices like Internet of Things (IoT) devices and microcontrollers (MCUs) are becoming increasingly common. These devices are more energyefficient and affordable, often with reduced features compared to the standard versions such as very limited memory and processing power for typical machine learning models. However, modern machine learning models can have millions of parameters, resulting in a large memory footprint. This complexity not only makes it difficult to deploy these large models on resource constrained devices but also increases the risk of latency and inefficiency in processing, which is crucial in some cases where real-time responses are required such as autonomous driving and medical diagnostics. In recent years, neural networks have seen significant advancements in model optimization techniques that help deployment and inference on these small devices. This narrative review offers a thorough examination of the progression and latest developments in neural network optimization, focusing on key areas such as quantization, pruning, knowledge distillation, and neural architecture search. It examines how these algorithmic solutions have progressed and how new approaches have improved upon the existing techniques making neural networks more efficient. This review is designed for machine learning researchers, practitioners, and engineers who may be unfamiliar with these methods but wish to explore the available techniques. It highlights ongoing research in optimizing networks for achieving better performance, lowering energy consumption, and enabling faster training times, all of which play an important role in the continued scalability of neural networks. Additionally, it identifies gaps in current research and provides a foundation for future studies, aiming to enhance the applicability and effectiveness of existing optimization strategies.

The article is dedicated to mathematic modeling of natural water desalination process by method of thermal distillation. The article gives the equations which allow describing the processes of film flowing and boiling of water, steam condensation and vacuum maintenance. The article presents the algorithm of calculation, implemented in MatLab computer mathematic system and Excel electronic tables, and the initial data required for the calculation. The model has been checked for adequacy. The calculation of ten-effect distillation system is given. The results of work can be used in design and optimization of process conditions for distillation systems.

The well-known “Sea battle” game is in the focus of the current job. The main goal of the article is to provide modified version of “Sea battle” game and to find optimal players’ strategies in the new rules. Changes were applied to attacking strategies (new option to attack hitting four cells in one shot was added) as well as to the size of the field (sizes of 10 × 10, 20 × 20, 30 × 30 were used) and to the rules of disposal algorithms during the game (new possibility to move the ship off the attacking zone). The game was solved with the use of game theory capabilities: payoff matrices were found for each version of altered rules, for which optimal pure and mixed strategies were discovered. For solving payoff matrices iterative method was used. The simulation was in applying five attacking algorithms and six disposal ones with parameters variation due to the game of players with each other. Attacking algorithms were varied in 100 sets of parameters, disposal algorithms — in 150 sets. Major result is that using such algorithms the modified “Sea battle” game can be solved — that implies the possibility of finding stable pure and mixed strategies of behaviour, which guarantee the sides gaining optimal results in game theory terms. Moreover, influence of modifying the rules of “Sea battle” game is estimated. Comparison with prior authors’ results on this topic was made. Based on matching the payoff matrices with the statistical analysis, completed earlier, it was found out that standard “Sea battle” game could be represented as a special case of game modifications, observed in this article. The job is important not only because of its applications in war area, but in civil areas as well. Use of article’s results could save resources in exploration, provide an advantage in war conflicts, defend devices under devastating impact.

А new cloud center of parallel computing is to be created in the Laboratory of Information Technologies (LIT) of the Joint Institute for Nuclear Research JINR) what is expected to improve significantly the efficiency of numerical calculations and expedite the receipt of new physically meaningful results due to the more rational use of computing resources. To optimize a scheme of parallel computations at a cloud environment it is necessary to test this scheme for various combinations of equipment parameters (processor speed and numbers, throughput оf а communication network etc). As a test problem, the parallel MPI algorithm for calculations of the long Josephson junctions (LDJ) is chosen. Problems of evaluating the impact of abovementioned factors of computing mean on the computing speed of the test problem are solved by simulation with the simulation program SyMSim developed in LIT.

The simulation of the LDJ calculations in the cloud environment enable users without a series of test to find the optimal number of CPUs with a certain type of network run the calculations in a real computer environment. This can save significant computational time in countable resources. The main parameters of the model were obtained from the results of the computational experiment conducted on a special cloud-based testbed. Computational experiments showed that the pure computation time decreases in inverse proportion to the number of processors, but depends significantly on network bandwidth. Comparison of results obtained empirically with the results of simulation showed that the simulation model correctly simulates the parallel calculations performed using the MPI-technology. Besides it confirms our recommendation: for fast calculations of this type it is needed to increase both, — the number of CPUs and the network throughput at the same time. The simulation results allow also to invent an empirical analytical formula expressing the dependence of calculation time by the number of processors for a fixed system configuration. The obtained formula can be applied to other similar studies, but requires additional tests to determine the values of variables.

The paper presents a mathematical model that describes the process of obtaining biogas from livestock waste. This model describes the processes occurring in a biogas plant for mesophilic and thermophilic media, as well as for continuous and periodic modes of substrate inflow. The values of the coefficients of this model found earlier for the periodic mode, obtained by solving the problem of model identification from experimental data using a genetic algorithm, are given.

For the model of methanogenesis, an optimal control problem is formulated in the form of a Lagrange problem, whose criterial functionality is the output of biogas over a certain period of time. The controlling parameter of the task is the rate of substrate entry into the biogas plant. An algorithm for solving this problem is proposed, based on the numerical implementation of the Pontryagin maximum principle. In this case, a hybrid genetic algorithm with an additional search in the vicinity of the best solution using the method of conjugate gradients was used as an optimization method. This numerical method for solving an optimal control problem is universal and applicable to a wide class of mathematical models.

In the course of the study, various modes of submission of the substrate to the digesters, temperature environments and types of raw materials were analyzed. It is shown that the rate of biogas production in the continuous feed mode is 1.4–1.9 times higher in the mesophilic medium (1.9–3.2 in the thermophilic medium) than in the periodic mode over the period of complete fermentation, which is associated with a higher feed rate of the substrate and a greater concentration of nutrients in the substrate. However, the yield of biogas during the period of complete fermentation with a periodic mode is twice as high as the output over the period of a complete change of the substrate in the methane tank at a continuous mode, which means incomplete processing of the substrate in the second case. The rate of biogas formation for a thermophilic medium in continuous mode and the optimal rate of supply of raw materials is three times higher than for a mesophilic medium. Comparison of biogas output for various types of raw materials shows that the highest biogas output is observed for waste poultry farms, the least — for cattle farms waste, which is associated with the nutrient content in a unit of substrate of each type.

This article reviews both historical achievements and modern results in the field of Markov Decision Process (MDP) and convex optimization. This review is the first attempt to cover the field of reinforcement learning in Russian in the context of convex optimization. The fundamental Bellman equation and the criteria of optimality of policy — strategies based on it, which make decisions based on the known state of the environment at the moment, are considered. The main iterative algorithms of policy optimization based on the solution of the Bellman equations are also considered. An important section of this article was the consideration of an alternative to the $Q$-learning approach — the method of direct maximization of the agent’s average reward for the chosen strategy from interaction with the environment. Thus, the solution of this convex optimization problem can be represented as a linear programming problem. The paper demonstrates how the convex optimization apparatus is used to solve the problem of Reinforcement Learning (RL). In particular, it is shown how the concept of strong duality allows us to naturally modify the formulation of the RL problem, showing the equivalence between maximizing the agent’s reward and finding his optimal strategy. The paper also discusses the complexity of MDP optimization with respect to the number of state–action–reward triples obtained as a result of interaction with the environment. The optimal limits of the MDP solution complexity are presented in the case of an ergodic process with an infinite horizon, as well as in the case of a non-stationary process with a finite horizon, which can be restarted several times in a row or immediately run in parallel in several threads. The review also reviews the latest results on reducing the gap between the lower and upper estimates of the complexity of MDP optimization with average remuneration (Averaged MDP, AMDP). In conclusion, the real-valued parametrization of agent policy and a class of gradient optimization methods through maximizing the $Q$-function of value are considered. In particular, a special class of MDPs with restrictions on the value of policy (Constrained Markov Decision Process, CMDP) is presented, for which a general direct-dual approach to optimization with strong duality is proposed.

Considered the application of physics-informed neural networks using multi layer perceptrons to solve Cauchy initial value problems in which the right-hand sides of the equation are continuous monotonically increasing, decreasing or oscillating functions. With the use of the computational experiments the influence of the construction of the approximate neural network solution, neural network structure, optimization algorithm and software implementation means on the learning process and the accuracy of the obtained solution is studied. The analysis of the efficiency of the most frequently used machine learning frameworks in software development with the programming languages Python and C# is carried out. It is shown that the use of C# language allows to reduce the time of neural networks training by 20–40%. The choice of different activation functions affects the learning process and the accuracy of the approximate solution. The most effective functions in the considered problems are sigmoid and hyperbolic tangent. The minimum of the loss function is achieved at the certain number of neurons of the hidden layer of a single-layer neural network for a fixed training time of the neural network model. It’s also mentioned that the complication of the network structure increasing the number of neurons does not improve the training results. At the same time, the size of the grid step between the points of the training sample, providing a minimum of the loss function, is almost the same for the considered Cauchy problems. Training single-layer neural networks, the Adam method and its modifications are the most effective to solve the optimization problems. Additionally, the application of twoand three-layer neural networks is considered. It is shown that in these cases it is reasonable to use the LBFGS algorithm, which, in comparison with the Adam method, in some cases requires much shorter training time achieving the same solution accuracy. The specificity of neural network training for Cauchy problems in which the solution is an oscillating function with monotonically decreasing amplitude is also investigated. For these problems, it is necessary to construct a neural network solution with variable weight coefficient rather than with constant one, which improves the solution in the grid cells located near by the end point of the solution interval.

The practical application of nuclear medicine demonstrates the continued information deficiency of the algorithms and programs that provide visualization and analysis of medical images. The aim of the study was to determine the principles of optimizing the processing of planar osteostsintigraphy on the basis of сomputer aided diagnosis (CAD) for analysis of texture descriptions of images of metastatic zones on planar scintigrams of skeleton. A computer-aided diagnosis system for analysis of skeletal metastases based on planar scintigraphy data has been developed. This system includes skeleton image segmentation, calculation of textural, histogram and morphometrical parameters and the creation of a training set. For study of metastatic images’ textural characteristics on planar scintigrams of skeleton was developed the computer program of automatic analysis of skeletal metastases is used from data of planar scintigraphy. Also expert evaluation was used to distinguishing ‘pathological’ (metastatic) from ‘physiological’ (non-metastatic) radiopharmaceutical hyperfixation zones in which Haralick’s textural features were determined: autocorrelation, contrast, ‘forth moment’ and heterogeneity. This program was established on the principles of сomputer aided diagnosis researches planar scintigrams of skeletal patients with metastatic breast cancer hearths hyperfixation of radiopharmaceuticals were identified. Calculated parameters were made such as brightness, smoothness, the third moment of brightness, brightness uniformity, entropy brightness. It has been established that in most areas of the skeleton of histogram values of parameters in pathologic hyperfixation of radiopharmaceuticals predominate over the same values in the physiological. Most often pathological hyperfixation of radiopharmaceuticals as the front and rear fixed scintigramms prevalence of brightness and smoothness of the image brightness in comparison with those of the physiological hyperfixation of radiopharmaceuticals. Separate figures histogram analysis can be used in specifying the diagnosis of metastases in the mathematical modeling and interpretation bone scintigraphy. Separate figures histogram analysis can be used in specifying the diagnosis of metastases in the mathematical modeling and interpretation bone scintigraphy.

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.

In the article, the questions of application of a optimization method for support of planning processes in warehouse systems by means of simulation are considered. Mechanisms of interrelation of optimization and simulation models are investigated, and also the algorithm of simulation optimization model development of warehouse system for support of planning processes is described in detail.