Small Stochastic Data Compactification Concept Justified in the Entropy Basis DOI Creative Commons
Viacheslav Kovtun, Elena Zaitseva, Vitaly Levashenko

et al.

Entropy, Journal Year: 2023, Volume and Issue: 25(12), P. 1567 - 1567

Published: Nov. 21, 2023

Measurement is a typical way of gathering information about an investigated object, generalized by finite set characteristic parameters. The result each iteration the measurement instance class object in form values An ordered instances forms collection whose dimensionality for real factor that cannot be ignored. Managing data collections, as well classification, regression, and clustering, are fundamental problems machine learning. Compactification approximation original equivalent (with reduced dimension parameters) with control accompanying capacity losses. Related to compactification completeness verifying procedure, which reliability assessment. If there stochastic parameters among initial parameters, procedure becomes more complicated. To take this into account, study proposes model structured defined terms relative entropy. such formalized iterative aimed at maximizing entropy sequential implementation direct reverse projections taking account estimates probability distribution densities their attributes. approximating function reduce computational complexity latter proposed. qualitatively assess undertakes formal analysis uses absolute share losses due compaction its metrics. Taking semantic connection completeness, proposed metric also relevant task assessing reliability. Testing proved both stability efficiency comparison previously used analogues, principal component method random projection method.

Language: Английский

A Comparative Assessment of the Cutting-Edge Time Series Forecasting Models Using Seasonal Data DOI
Kyrylo Yemets, Nataliia Lotoshynska

Lecture notes on data engineering and communications technologies, Journal Year: 2025, Volume and Issue: unknown, P. 206 - 217

Published: Jan. 1, 2025

Language: Английский

Citations

0
An ensemble approach integrating LSTM and ARIMA models for enhanced financial market predictions DOI Creative Commons
Lesia Mochurad,

Andrii Dereviannyi

Royal Society Open Science, Journal Year: 2024, Volume and Issue: 11(9)

Published: Sept. 1, 2024

Forecasting financial markets is a complex task that requires addressing various challenges, such as market complexity, data heterogeneity, the need for rapid response and constant changes in conditions, to gain competitive advantage. To effectively address these it necessary constantly improve existing develop new methods of intelligent forecasting, which will accuracy forecasts, reduce risks increase productivity decision-making processes. In this article, we study analyse forecasting markets, support vector regression (SVR), autoregressive integrated moving average (ARIMA), long short-term memory recurrent neural network (LSTM) extreme gradient boosting algorithm (XG-Boost). Based on analysis, propose an ensemble procedure integrates LSTM ARIMA models. Due careful combination models, our approach yields better results than individual methods. For example, model demonstrates significant 15% improvement root mean square error (RMSE) slight coefficient determination compared with LSTM. Furthermore, simulation obtained three real-world datasets evaluated using RMSE criterion confirm superiority proposed method over alternative LSTMs, transformer models optimized deep networks forecasting. creates prerequisites parallelizing both thus providing opportunity accelerate future research.

Language: Английский

Citations

3
Investigation of the Information Interaction of the Sensor Network End IoT Device and the Hub at the Transport Protocol Level DOI Open Access
Viacheslav Kovtun, Krzysztof Grochla, Konrad Połys

et al.

Electronics, Journal Year: 2023, Volume and Issue: 12(22), P. 4662 - 4662

Published: Nov. 15, 2023

The study examines the process of information transfer between sensor network end IoT device and hub at transport protocol level focused on using 5G platform. authors interpreted researched as a semi-Markov (focused dynamics size sliding window) with two nested Markov chains (the first characterizes current window, second, number data blocks sent value this characteristic). As result, stationary distribution window was obtained both for resulting chains, etc. A recursive approach to calculation mentioned is formalized. This characterized by linear computational complexity. Based function formulated that bandwidth communication channel entities specified in research object. Using mathematical apparatus, Window Scale parameter TCP Westwood+ tuned. Testing has shown superiority modified over basic versions BIC TCP, Vegas, NewReno, Veno protocols conditions points wireless environment.

Language: Английский

Citations

6
Evaluation of the QoS policy model of an ordinary 5G smart city cluster with predominant URLLC and eMBB traffic DOI Creative Commons
Viacheslav Kovtun, Krzysztof Grochla, Torki Altameem

et al.

PLoS ONE, Journal Year: 2023, Volume and Issue: 18(12), P. e0295252 - e0295252

Published: Dec. 8, 2023

A typical element of the smart city’s information and communication space is a 5G cluster, which focused on serving both new handover requests because it an open system. In ordinary city Ultra-Reliable Low-Latency Communications (URLLC) enhanced Mobile BroadBand (eMBB) traffic types prevail. The formation effective QoS policy for such object (taking into account potentially active slicing technology) urgent problem. As baseline, this research considers Quality Service (QoS) with constraints context-defined URLLC eMBB classes incoming requests. Evaluating instance defined within framework basic concept requires formalization complete qualitative metric computationally efficient mathematical apparatus its calculation. article presents accurate approximate methods calculating quality parameters as probability loss typed utilization ratio resource, depend implementation estimated policy. At same time, original parametric includes fixed characteristics (amount available resources, load according to request classes) controlled due specifics concept. paper empirically proves adequacy presented evaluating scope research. Also, in proposed metric, comparison author’s parametrically close analogue (the well-known scheme, takes phenomenon reservation resources), determined taking was made. results testify favour superiority approach metrics.

Language: Английский

Citations

4
Potential Breakthroughs in Environmental Monitoring and Management DOI
Abdulhalim Musa Abubakar, Irnis Azura Zakarya,

Mohammad Hasnain

et al.

Advances in geospatial technologies book series, Journal Year: 2024, Volume and Issue: unknown, P. 239 - 282

Published: Dec. 6, 2024

Environmental monitoring and management are critical for sustainable development the preservation of natural resources. Recent advancements in artificial intelligence (AI) integrated with geospatial technologies promise to revolutionize these fields. Delving into potential breakthroughs, AI offers precise, real-time environmental parameters through machine learning (ML) algorithms, remote sensing data, geographic information systems (GIS). Enhanced data analysis techniques facilitate early detection anomalies, predictive modeling ecological trends, efficient resource management. Successful implementations tracking climate change impacts, managing disasters, biodiversity presented various case studies. Challenges such as privacy, algorithm transparency, need interdisciplinary collaboration also addressed. Future research directions explore AI's foster more resilient adaptive practices. Synthesizing technology underscore a transformative approach safeguarding our environment.

Language: Английский

Citations

1
The Markov Concept of the Energy Efficiency Assessment of the Edge Computing Infrastructure Peripheral Server Functioning over Time DOI Open Access
Viacheslav Kovtun, Torki Altameem, Mohammed Al‐Maitah

et al.

Electronics, Journal Year: 2023, Volume and Issue: 12(20), P. 4320 - 4320

Published: Oct. 18, 2023

The article is devoted to the research of peripheral server energy consumption managing process defined based on threshold policy by manipulating values characteristic parameters arithmetic-logical complex latter. object formalized a Markov queue model with single-threshold control scheme for intensity accepted requests service. A feature life cycle non-stationary mode operation in terms consumption, due need periodically poll controlled sensor network and received data real-time. To take into account this circumstance, intensities transitions heterogeneous birth-and-death created are interpreted as non-random periodic functions time. resulting mathematical apparatus summarized metric that allows us estimate target server’s both at specific moment censored time interval (this distinguishes obtained result from existing analogs).

Language: Английский

Citations

3
Small Stochastic Data Compactification Concept Justified in the Entropy Basis DOI Creative Commons
Viacheslav Kovtun, Elena Zaitseva, Vitaly Levashenko

et al.

Entropy, Journal Year: 2023, Volume and Issue: 25(12), P. 1567 - 1567

Published: Nov. 21, 2023

Measurement is a typical way of gathering information about an investigated object, generalized by finite set characteristic parameters. The result each iteration the measurement instance class object in form values An ordered instances forms collection whose dimensionality for real factor that cannot be ignored. Managing data collections, as well classification, regression, and clustering, are fundamental problems machine learning. Compactification approximation original equivalent (with reduced dimension parameters) with control accompanying capacity losses. Related to compactification completeness verifying procedure, which reliability assessment. If there stochastic parameters among initial parameters, procedure becomes more complicated. To take this into account, study proposes model structured defined terms relative entropy. such formalized iterative aimed at maximizing entropy sequential implementation direct reverse projections taking account estimates probability distribution densities their attributes. approximating function reduce computational complexity latter proposed. qualitatively assess undertakes formal analysis uses absolute share losses due compaction its metrics. Taking semantic connection completeness, proposed metric also relevant task assessing reliability. Testing proved both stability efficiency comparison previously used analogues, principal component method random projection method.

Language: Английский

Citations

2