The object of research is multi-agent systems based on Deep Reinforcement Learning algorithms and analysis of ways to establish interaction within the system, based on intelligent agents. Also, part of the material in this paper covers ways to organize the management and administration of agents at the meta-level: external controllers and tools to optimize their work, describing architectural solutions that should accelerate agents' training. The studied full-fledged multi-agent system would be flexible to expansion and would give effective acceleration in agent training and problem-solving quality.In this paper, the following neural network models were considered: DQN, DDQN, PPO, TD (methods based on Q-Learning), an approach using a neural network with Monte-Carlo tree search. The presented models were tested on a Sudoku problem with a dataset of 5039 combinations, dimensions 2 × 2, 4 × 4, and 9 × 9. Several sets of agent rewards were used. The presentation of data during the learning and problem-solving process was described. Also was built a multi-agent system based on the model using a Monte-Carlo tree search.According to the study results, it was revealed that for tasks in a complex environment, the models based on Q-Learning are practically ineffective (plots support the statement). The training process for these models is quite demanding on the characteristics of the workstation hardware. It was also determined that the Monte-Carlo tree search method does a good job. Even with a small number of iterations, it shows results better than other Deep Learning methods (45-50 % accuracy for 9 × 9). However, a significant drawback is a complexity of training the model, and the hardware requirements are too large for this kind of research.

The object of research is the use of multi-agent systems for text data mining. The need for this study arose with a tendency to increase the amount of textual information gene rated in the world. Accordingly, it is necessary to develop and research methods of its processing, as well as ways to use the results of this processing, because the methods can't exist in isolation from practice. At the same time, there is a development of multi-agent systems (MAS), where agents are endowed with some kind of intelligence, these systems can be easily scaled. The use of MAS for text analysis is a promising area. The following methods of text data analysis were used in this study: TF-IDF and RAKE methods, Word2Vec neural network models, and TextRank. The algorithms were compared for their work and the results were compared. The corpus of documents (10-12 texts, 5732-12331 words) from the subject areas of physics and biology were used as a test set. According to the results of the study, one method was chosen, on the basis of which the MAS was built to solve the problem. Additionally, Schulze methods (with one and several winners) were used for voting. With the received system additional researches concerning accuracy and speed of work, and also-influence are carried out system parameters for its operation. It has been found that TF-IDF-based analysis is useful for finding terms in documents with a weak context. The resulting system shows an accuracy of 75 % (3 of the 4 words proposed by the system are terms). The maximum operating time on test cases is 2-3 seconds, which is achieved through the use of parallel calculations and modification of the Schulze method. The results obtained in this paper are heuristic (ontology is a rather vague concept) and require additional elaboration by experts in the relevant fields. However, the results are positive within this experiment.

The object of research is the methods of fast classification for solving text data classification problems. The need for this study is due to the rapid growth of textual data, both in digital and printed forms. Thus, there is a need to process such data using software, since human resources are not able to process such an amount of data in full. A large number of data classification approaches have been developed. The conducted research is based on the application of the following methods of classification of text data: Bloom filter, naive Bayesian classifier and neural networks to a set of text data in order to classify them into categories. Each method has both disadvantages and advantages. This paper will reflect the strengths and weaknesses of each method on a specific example. These algorithms were comparatively among themselves in terms of speed and efficiency, that is, the accuracy of determining the belonging of a text to a certain class of classification. The work of each method was considered on the same data sets with a change in the amount of training and test data, as well as with a change in the number of classification groups. The dataset used contains the following classes: world, business, sports, and science and technology. In real conditions of the classification of such data, the number of categories is much larger than that considered in the work, and may have subcategories in its composition. In the course of this study, each method was analyzed using different parameter values to obtain the best result. Analyzing the results obtained, the best results for the classification of text data were obtained using a neural network.

No abstract

Національний технічний університет України «Київський політехнічний інститут імені Ігоря Сікорського» Тарасенко М.В. Національний технічний університет України «Київський політехнічний інститут імені Ігоря Сікорського» ПОРІВНяльНИй АНАлІЗ ПРОгРАМНИх БІБлІОТЕК для КлАСИфІКАцІЇ ТЕКСТОВИх дАНИх ІЗ ВИКОРИСТАННяМ ШТУЧНИх НЕйРОННИх МЕРЕЖ У цій роботі розглянуті бібліотеки для вирішення задач машинного навчання. Виконанно порівняльний аналіз даних бібліотек. Для порівняння було обрано задачу класифікації текстових даних. Навчання моделі відбувалось методами навчання з вчителем. Для навчання були використані штучні нейронні мережі з нейронами, що мають довгу короткочасну пам'ять. Оцінка бібліотек відбувається за точністю класифікації, швидкістю навчання моделі в однакових середовищах та наявністю засобів, що полегшують побудову та апробацію цієї моделі. Ключові слова: штучні нейронні мережі, глибоке навчання, інтелектуальний аналіз текстових даних, класифікація текстів, навчання з вчителем. СРАВНИТЕльНЫй АНАлИЗ ПРОгРАММНЫх БИБлИОТЕК для КлАССИфИКАцИИ ТЕКСТОВЫх дАННЫх С ИСПОльЗОВАНИЕМ ИСКУССТВЕННЫх НЕйРОННЫх СЕТЕй В работе рассмотрены библиотеки для решения задач машинного обучения. Выполнен сравнительный анализ данных библиотек. Для сравнения были выбраны задачу классификации текстовых данных. Обучение модели происходило методами обучения с учителем. Для обучения были использованы искусственные нейронные сети с нейронами, имеющими долгую кратковременную память. Оценка библиотек происходит по точности классификации, скорости обучения модели в одинаковых средах и наличию средств, облегчающих построение и апробацию данной модели. Ключевые слова: искусственные нейронные сети, глубокое обучение, интеллектуальный анализ текстовых данных, классификация текстов, обучение с учителем. COMPArATIVE ANALYSIS OF SOFTWArE LIBrArIES FOr ThE CLASSIFICATION OF TEXT DATA uSING ArTIFICIAL NEurAL NETWOrKSIn this paper libraries for solving machine learning problems are being considered. Comparative analysis of library data is being performed. For comparison, the task of text data classification was selected. The model was taught using supervised learning methods. For training, artificial neural networks with neurons with long short-term memory were used. The evaluation of libraries is based on the accuracy of the classification, the speed of learning the model in the same environment and the availability of tools that facilitate the construction and testing of machine learning model.