Viktor V. Semenov

Viktor V. Semenov

Auto-generated profile, based on publicly available information

Viktor V. Semenov

St. Petersburg Federal Research Center of the Russian Academy of Sciences

Research Focus & Metrics

Academic Career Length

32Years (1995-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2019 in Journal of Food Composition and Analysis
Verified
Journal Article
Closed access
VSViktor Semenov
Viktor Semenov
SVSergey Volkov
Sergey Volkov
MKMaria Khaydukova
Maria Khaydukova
AFAlexander Fedorov
Alexander Fedorov
ILIrina Lisitsyna
Irina Lisitsyna
ALAndrey Legin
Andrey Legin

1 November 2009 in Doklady Earth Sciences
Verified
Journal Article
Closed access
PDF file
Andrei VelichkoAndrei Velichko
Andrei Velichko
NCN. Catto
N. Catto
ATAlexey S. Tesakov
Alexey S. Tesakov
VTVadim Titov
Vadim Titov
TMT. D. Morozova
T. D. Morozova
STS. N. Timireva
S. N. Timireva

1 October 2025 in Emerging Science Journal
Verified
Journal Article
Open access
PDF file
ILIlya S Lebedev
Ilya S Lebedev
MSMikhail Sukhoparov
Mikhail Sukhoparov
VSViktor Semenov
Viktor Semenov
DKDmitry Khasanov
Dmitry Khasanov
The research objective is to construct an adaptive model for modular machine learning structures that improves the processing quality of information sequences. The novelty of the proposed methodology is that it can identify segments of an information sequence obtained using various methods and assign models with the best quality indicator values to subsequences. Classical methods allow tuning of the model to the entire data sample. The improvement consists of the proposed solutions that consider the inverse problem of forming segments of data sequences, such that their properties correspond to the processing model. The proposed methodology was tested on various models and datasets. Segmentation and assignment of regression models with the best characteristics to individual segments allow the reduction of the mean square error (MSE) and mean absolute error (MAE) to 8%. The findings show an opportunity to increase of 5-8% for weak LR, SVM, and GR models, while strong DT, CNN, ANN, ANFIS, and XGBoost models improve by 1-4% in scenarios with limited data. Segmentation enables more efficient training and reduces sensitivity to noise and outliers. The proposed solution allows the selection of segmentation strategies and model combinations considering local data properties. Its application enables the implementation of existing machine learning architectures to improve the quality of training and analysis of information sequences and increase adaptability, scalability, and interpretability.

12 September 2019
Verified
Book Chapter
Closed access
PDF file
VSViktor V. Semenov
Viktor V. Semenov
ILIlya S Lebedev
Ilya S Lebedev
MSMikhail E. Sukhoparov
Mikhail E. Sukhoparov
KSKseniya I. Salakhutdinova
Kseniya I. Salakhutdinova


1 December 2022 in Scientific and Technical Journal of Information Technologies, Mechanics and Optics
Verified
Journal Article
Open access
PDF file
VSV.V. Semenov
V.V. Semenov
The wide spread of cyber-physical systems, as well as the widespread integration of computing resources into physical entities, have led to an increase in the risks of deliberate and accidental security incidents. In this regard, the development of new methods and tools and improvement of the existing ones for monitoring such systems is of particular relevance. The methods being created and modernized should have increased recall and precision of identification, especially for critical infrastructure objects. An original method for processing data for monitoring the state of cyber-physical systems based on time series analysis using significance weights as a post-processing of classification results was proposed. The method differs from the existing ones by the combined approach that combines the use events of information security and functional safety in monitoring systems. It is characterized by the use of an ensemble of decision trees as well as parallel classifiers and Fishburn weight coefficients in the analysis of the set of the most informative features obtained from time series. The applicability of the method was substantiated by conducting of a computational experiment on a known data set which characterizes the functioning of the information and physical components in the implementation of various types of attacks on the components of the experimental stand of the cyber-physical water treatment system. When using the developed method, the identification precision increased by 1.45 % compared to the best approaches presented in other scientific works, and the recall increased by 4.45 % and amounted to 99.85 % for both indicators. The results obtained are adapted for practical use in systems for identifying the state of cyber-physical systems. The theoretical significance lies in the possibility of using the results of the study in the design of systems for monitoring the information security and functional safety of cyber-physical systems.