All issues

The probabilistic topic model of a text document collection finds two matrices: a matrix of conditional probabilities of topics in documents and a matrix of conditional probabilities of words in topics. Each document is represented by a multiset of words also called the “bag of words”, thus assuming that the order of words is not important for revealing the latent topics of the document. Under this assumption, the problem is reduced to a low-rank non-negative matrix factorization governed by likelihood maximization. In general, this problem is ill-posed having an infinite set of solutions. In order to regularize the solution, a weighted sum of optimization criteria is added to the log-likelihood. When modeling large text collections, storing the first matrix seems to be impractical, since its size is proportional to the number of documents in the collection. At the same time, the topical vector representation (embedding) of documents is necessary for solving many text analysis tasks, such as information retrieval, clustering, classification, and summarization of texts. In practice, the topical embedding is calculated for a document “on-the-fly”, which may require dozens of iterations over all the words of the document. In this paper, we propose a way to calculate a topical embedding quickly, by one pass over document words. For this, an additional constraint is introduced into the model in the form of an equation, which calculates the first matrix from the second one in linear time. Although formally this constraint is not an optimization criterion, in fact it plays the role of a regularizer and can be used in combination with other regularizers within the additive regularization framework ARTM. Experiments on three text collections have shown that the proposed method improves the model in terms of sparseness, difference, logLift and coherence measures of topic quality. The open source libraries BigARTM and TopicNet were used for the experiments.

In the first part of the paper we present convergence analysis of AGMsDR method on a new class of functions — in general non-convex with $M$-Lipschitz-continuous gradients that satisfy Polyak – Lojasiewicz condition. Method does not need the value of $\mu^{PL}>0$ in the condition and converges linearly with a scale factor $\left(1 - \frac{\mu^{PL}}{M}\right)$. It was previously proved that method converges as $O\left(\frac1{k^2}\right)$ if a function is convex and has $M$-Lipschitz-continuous gradient and converges linearly with a~scale factor $\left(1 - \sqrt{\frac{\mu^{SC}}{M}}\right)$ if the value of strong convexity parameter $\mu^{SC}>0$ is known. The novelty is that one can save linear convergence if $\frac{\mu^{PL}}{\mu^{SC}}$ is not known, but without square root in the scale factor.

The second part presents modification of AGMsDR method for solving problems that allow alternating minimization (Alternating AGMsDR). The similar results are proved.

As the result, we present adaptive accelerated methods that converge as $O\left(\min\left\lbrace\frac{M}{k^2},\,\left(1-{\frac{\mu^{PL}}{M}}\right)^{(k-1)}\right\rbrace\right)$ on a class of convex functions with $M$-Lipschitz-continuous gradient that satisfy Polyak – Lojasiewicz condition. Algorithms do not need values of $M$ and $\mu^{PL}$. If Polyak – Lojasiewicz condition does not hold, the convergence is $O\left(\frac1{k^2}\right)$, but no tuning needed.

We also consider the adaptive catalyst envelope of non-accelerated gradient methods. The envelope allows acceleration up to $O\left(\frac1{k^2}\right)$. We present numerical comparison of non-accelerated adaptive gradient descent which is accelerated using adaptive catalyst envelope with AGMsDR, Alternating AGMsDR, APDAGD (Adaptive Primal-Dual Accelerated Gradient Descent) and Sinkhorn's algorithm on the problem dual to the optimal transport problem.

Conducted experiments show faster convergence of alternating AGMsDR in comparison with described catalyst approach and AGMsDR, despite the same asymptotic rate $O\left(\frac1{k^2}\right)$. Such behavior can be explained by linear convergence of AGMsDR method and was tested on quadratic functions. Alternating AGMsDR demonstrated better performance in comparison with AGMsDR.

The technique for solving the logistic task of fuel supply in the region, including the interconnected tasks of routing, clustering, optimal distribution of resources and stock control is proposed. The calculations have been carried out on the example of fuel supply system of the Udmurt Republic.

The purpose of this study is to develop a methodology for change points detection in time series, including financial data. The theoretical basis of the study is based on the pieces of research devoted to the analysis of structural changes in financial markets, description of the proposed algorithms for detecting change points and peculiarities of building classical and deep machine learning models for solving this type of problems. The development of such tools is of interest to investors and other stakeholders, providing them with additional approaches to the effective analysis of financial markets and interpretation of available data.

To address the research objective, a neural network was trained. In the course of the study several ways of training sample formation were considered, differing in the nature of statistical parameters. In order to improve the quality of training and obtain more accurate results, a methodology for feature generation was developed for the formation of features that serve as input data for the neural network. These features, in turn, were derived from an analysis of mathematical expectations and standard deviations of time series data over specific intervals. The potential for combining these features to achieve more stable results is also under investigation.

The results of model experiments were analyzed to compare the effectiveness of the proposed model with other existing changepoint detection algorithms that have gained widespread usage in practical applications. A specially generated dataset, developed using proprietary methods, was utilized as both training and testing data. Furthermore, the model, trained on various features, was tested on daily data from the S&P 500 index to assess its effectiveness in a real financial context.

As the principles of the model’s operation are described, possibilities for its further improvement are considered, including the modernization of the proposed model’s structure, optimization of training data generation, and feature formation. Additionally, the authors are tasked with advancing existing concepts for real-time changepoint detection.

The problem of centralized planning for data transmission routes in delay tolerant networks is considered. The original problem is extended with additional requirements to nodes storage and communication process. First, it is assumed that the connection between the nodes of the graph is established using antennas. Second, it is assumed that each node has a storage of finite capacity. The existing works do not consider these requirements. It is assumed that we have in advance information about messages to be processed, information about the network configuration at specified time points taken with a certain time periods, information on time delays for the orientation of the antennas for data transmission and restrictions on the amount of data storage on each satellite of the grouping. Two wellknown algorithms — CGR and Earliest Delivery with All Queues are improved to satisfy the extended requirements. The obtained algorithms solve the optimal message routing problem separately for each message. The problem of validation of the algorithms under conditions of lack of test data is considered as well. Possible approaches to the validation based on qualitative conjectures are proposed and tested, and experiment results are described. A performance comparison of the two implementations of the problem solving algorithms is made. Two algorithms named RDTNAS-CG and RDTNAS-AQ have been developed based on the CGR and Earliest Delivery with All Queues algorithms, respectively. The original algorithms have been significantly expanded and an augmented implementation has been developed. Validation experiments were carried to check the minimum «quality» requirements for the correctness of the algorithms. Comparative analysis of the performance of the two algorithms showed that the RDTNAS-AQ algorithm is several orders of magnitude faster than RDTNAS-CG.

We propose a family of Gauss –Newton methods for solving optimization problems and systems of nonlinear equations based on the ideas of using the upper estimate of the norm of the residual of the system of nonlinear equations and quadratic regularization. The paper presents a development of the «Three Squares Method» scheme with the addition of a momentum term to the update rule of the sought parameters in the problem to be solved. The resulting scheme has several remarkable properties. First, the paper algorithmically describes a whole parametric family of methods that minimize functionals of a special kind: compositions of the residual of a nonlinear equation and an unimodal functional. Such a functional, entirely consistent with the «gray box» paradigm in the problem description, combines a large number of solvable problems related to applications in machine learning, with the regression problems. Secondly, the obtained family of methods is described as a generalization of several forms of the Levenberg –Marquardt algorithm, allowing implementation in non-Euclidean spaces as well. The algorithm describing the parametric family of Gauss –Newton methods uses an iterative procedure that performs an inexact parametrized proximal mapping and shift using a momentum term. The paper contains a detailed analysis of the efficiency of the proposed family of Gauss – Newton methods; the derived estimates take into account the number of external iterations of the algorithm for solving the main problem, the accuracy and computational complexity of the local model representation and oracle computation. Sublinear and linear convergence conditions based on the Polak – Lojasiewicz inequality are derived for the family of methods. In both observed convergence regimes, the Lipschitz property of the residual of the nonlinear system of equations is locally assumed. In addition to the theoretical analysis of the scheme, the paper studies the issues of its practical implementation. In particular, in the experiments conducted for the suboptimal step, the schemes of effective calculation of the approximation of the best step are given, which makes it possible to improve the convergence of the method in practice in comparison with the original «Three Square Method». The proposed scheme combines several existing and frequently used in practice modifications of the Gauss –Newton method, in addition, the paper proposes a monotone momentum modification of the family of developed methods, which does not slow down the search for a solution in the worst case and demonstrates in practice an improvement in the convergence of the method.

The paper presents a software system aimed at finding best (in some sense) parameters of an algorithm. The system handles both discrete and continuous parameters and employs massive parallelism offered by public clouds. The paper presents an overview of the system, a method to measure algorithm's performance in the cloud and numerical results of system's use on several problem sets.

The decomposition algorithms provide approaches to deal with NP-hardness in solving discrete optimization problems (DOPs). In this article one of the promising ways to exploit sparse matrices — local elimination algorithm in parallel interpretation (LEAP) are demonstrated. That is a graph-based structural decomposition algorithm, which allows to compute a solution in stages such that each of them uses results from previous stages. At the same time LEAP heavily depends on elimination ordering which actually provides solving stages. Also paper considers tree- and block-parallel for LEAP and required realization process of it comparison of a several heuristics for obtaining a better elimination order and shows how is related graph structure, elimination ordering and solving time.

The paper presents an implementation of branch and bound algorithm employing coarse grained parallelism. The system is based on CBC (COIN-OR branch and cut) open-source MIP solver and inter-process communication capabilities of Erlang. Numerical results show noticeable speedup in comparison to single-threaded CBC instance.

Molecular dynamic methods that use ReaxFF force field allow one to obtain sufficiently good results in simulating large multicomponent chemically reactive systems. Here is represented an algorithm of searching optimal parameters of molecular-dynamic force field ReaxFF for arbitrary chemical systems and its implementation. The method is based on the multidimensional technique of global minimum search suggested by R.G. Strongin. It has good scalability useful for running on distributed parallel computers.