Перегляд за Автор "Krepych, Svitlana"
Зараз показуємо 1 - 3 з 3
Результатів на сторінці
Налаштування сортування
Документ Forecasting system of utilities service costs based on neural network(Національний технічний університет "Харківський політехнічний інститут", 2020) Krepych, Svitlana; Spivak, IrynaThe work is devoted to the problem of excessive spending of people's funds on utilities, especially in winter, when these costs can amount to more than 25% of the family budget. The question of the possibility of saving at least part of these costs by monitoring their possible value and reducing this indicator is an urgent task. Hence, the development of a software system for forecasting utility costs is an urgent practical task. To solve this problem, the authors propose to use a neural network, because it is advisable to use it in situations where there is predetermined known information and on its basis the user needs to get the predicted new information. The method for forecasting utility costs based on the use of a neural network takes into account user's data of utility service costs entered manually or obtained from the EPS system, which is convenient because you can immediately get a large set of input data to more accurately predict future costs. Another type of input data is data obtained from weather forecast sites to determine forecast indicators for correct the training of neural network. Based on these data, the network studies and builds a separate model for forecasting utility costs for each user. Considering that the data on utility service costs entered by users into the system each month may not match the date, it is proposed to take into account this in accuracy, to given the input data for forecasting as an interval corridor of values which containing the minimum and maximum forecast limits. The developed software system and the method of forecasting utility service costs were tested on the example of a real user of the EPS system.Документ Hybrid approach for data filtering and machine learning inside content management system(Національний технічний університет "Харківський політехнічний інститут", 2023) Poliarush, Oleh; Krepych, Svitlana; Spivak, IrynaThe object of research is the processes of data filtering and machine learning in content management systems. The subject of research is developing a hybrid approach to data filtering based on a combination of supervised and unsupervised machine learning. The article explores machine learning approaches to content management and how they can change the way we organize, categorize, and derive value from vast amounts of data. The main goal is to develop and use a hybrid approach for data filtering and training that will help optimize resource consumption and perform supervised training for better categorization in the future. This approach includes elements of supervised and unsupervised learning using the BERT architecture that uses this kind of flow that help reduce resource usage and adjust the algorithm to perform better in a specific area. As a result, thanks to this approach, the intelligent system was able to independently optimize for a specific field of use and help to reduce the costs of using resources. Conclusion. After applying a hybrid approach of data filtering and machine learning to existing data streams, we obtain a performance increase of up to 5%, and this percentage increases depending on the running time of the application.Документ Improvement of SVD algorithm to increase the efficiency of recommendation systems(Національний технічний університет "Харківський політехнічний інститут", 2021) Krepych, Svitlana; Spivak, IrynaMany existing websites use recommendation systems for their users. They generate various offers for them, for example, similar products or recommend the people registered on this site with similar interests. Such referral mechanisms process vast amounts of information to identify potential user preferences. Recommendation systems are programs that try to determine what users want to find, what might interest them, and recommend it to them. These mechanisms have improved the interaction between the user and the site. Instead of static information, they provide dynamic information that changes: recommendations are generated separately for each user, based on his previous activity on this web resource. Information from other visitors may also be taken into account. The methods of collecting information provided by the Internet have greatly simplified the use of human thought through collaborative filtering. But, on the other hand, the large amount of information complicates the implementation of this possibility. For example, the behavior of some people is quite clearly amenable to modeling, while others behave completely unpredictably. And it is the latter that affect the shift of the results of the recommendation system and reduce its effectiveness. An analysis of Internet resources has shown that most of the recommendation systems do not provide recommendations to users, and the part that does, for example, offers products to the user, selects recommendations manually. Therefore, the task of developing methods for automated generation of recommendations for a limited set of input data is quite relevant. The problems of data sparseness, new user problem, scalability of the widely used SVD algorithm for the development of such recommendation systems are proposed to be eliminated by improving this algorithm by the method of the nearest k-neighbors. This method will allow you to easily segment and cluster system data, which will save system resources.