Vol 14, No 1 (2023)
Tabular information recognition using convolutional neural networks
Abstract
The relevance of identifying tabular information and recognizing its contents for processing scanned documents is shown. The formation of a data set for training, validation and testing of a deep learning neural network (DNN) YOLOv5s for the detection of simple tables is described. The effectiveness of using this DNN when working with scanned documents is shown. Using the Keras Functional API, a convolutional neural network (CNN) was formed to recognize the main elements of tabular information — numbers, basic punctuation marks and Cyrillic letters. The results of a study of the work of this CNN are given. The implementation of the identification and recognition of tabular information on scanned documents in the developed IS updating information in databases for the Unified State Register of Real Estate system is described.
Program Systems: Theory and Applications. 2023;14(1):3-30
3-30
Decomposition of construction method for a language encoder
Abstract
An encoder as part of a language model is a mechanism for converting text information into an effective numerical representation which is suitable for solving a wide range of text processing tasks by means of neural network methods. This paper suggests a way of decomposing of the learning process for a language encoder. The author considers the issues of expediency of such decomposition taking into account reduction of computational costs, quality control at intermediate training stages, provision of the interpretability of the results on each stage. The quality evaluation of the encoder is given.
Program Systems: Theory and Applications. 2023;14(1):31-54
31-54
Steady-state performance analysis of multiserver queueing models with redundancy
Abstract
An approach to study of the stochastic models of distributed computing systems by means of the multiserver queueing models with redundancy is suggested. Perspectives and limitations of the approach, as well as possible future research directions are presented.
Program Systems: Theory and Applications. 2023;14(1):55-94
55-94
A system for extracting symptom mentions from texts by means of neural networks
Abstract
This paper presents a system for extracting symptom mentions from medical texts in natural (Russian) language. The system finds symptom mentions in texts, brings them to a standard form and identifies the found symptom to a group of similar symptoms. For each stage of processing we use a separate neural network. We extract symptoms of three areas of diseases: allergic and pulmonological diseases, as well as coronavirus infection (COVID-19). We present and describe an annotated corpus of sentences that is used to train neural networks for extracting symptom mentions. These sentences were marked up with the help of a simple XML-like language. An extended BIO-markup format was proposed for the sentences directly received at the input of the neural network. We give the quality evaluation of the symptom extraction accuracy under strict and flexible testing. Possible approaches to normalization and identification of symptom mentions and their implementation are described. Our results are compared with those achieved in similar researches, thus we show the place of our system among clinical decision support systems.
Program Systems: Theory and Applications. 2023;14(1):95-123
95-123
About One Class of Discrete-Continuous Systems with Parameters
Abstract
The study focuses on a special case of a hybrid system: discrete-continuous systems (DCS) with parameters and intermediate criteria. Such systems are two-level. The parameters are included only in continuous systems operating alternately at the lower level. The upper level is described by a discrete process and plays a connecting role for all the lower-level systems. The upper level also determines the policy of interaction of lower-level systems and provides minimization of functionality. The authors formulate an analogue of sufficient Krotov optimality conditions and construct a method for improving control and parameters. The paper contains an illustrative example. Based on the general conditions obtained, we have researched a special case: quasilinear DNS.
Program Systems: Theory and Applications. 2023;14(1):125-148
125-148