Selection of network infrastructure monitoring parameters to classify network status
Article Sidebar
Main Article Content
Abstract
The subject of research in the article is the stage of preliminary data processing for machine learning algorithms and consideration of various pre-processing techniques and evaluation the informativeness of features-based parameters network infrastructure monitoring for effective intellectual state analysis. The aim of the work - to consider various data preprocessing techniques and evaluation of informativeness for determining controls parameters of network infrastructure for more efficient intellectual analysis. The article solves following tasks: consideration of methods for selecting parameters; parameter determination for assessing the state of a network filtration methods, based on algorithms that are not related to classification methods; wrapper methods, based on importance features information, obtained from classification or regression methods, which can determine data deeper patterns; embedded methods that perform feature selection during the classifier training procedure and optimize the set of features used to improve accuracy. Results: various preliminary processing techniques and evaluation of informativeness of feature were analyzed to determine the parameters of network infrastructure monitoring. The results of feature selection methods were analyzed to simplify the different machine learning models. The minimum parameters set has been formed for monitoring the state of the network infrastructure. Conclusions: The use of feature selection methods made it possible to reduce the input parameter set for classifying the state of the network infrastructure methods.
Article Details
References
Ruban, I., Martovytskyi, V. and Lukova-Chuiko, N. (2016), “Designing a monitoring model for cluster super–computer“, East-ern-European Journal of Enterprise Technologies, Vol. 6, No. 84, pp. 32-37.
Katasеv, A.S., Katasеva, D.V. and Kyrpychnykov, A.P. (2015), “Neural network diagnosis of abnormal network activity“, Bulletin of Kazan Technological University, No. 18, pp. 163-167.
Liu L. (2014), “Anomaly diagnosis based on regression and classification analysis of statistical traffic features”, Security and Communication Networks, No. 7, pp. 132-138.
Gavalas D. (2000), “Advanced network monitoring applications based on mobile/intelligent agent technology“, Computer Communications, No. 23, pp 720-730, DOI: https://doi.org/10.1016/S0140-3664(99)00232-7.
Ruban, I., Martovytskyi V. and Lukova-Chuiko N. (2016), “An approach to classifying a network state based on statistical parameters for detecting anomalies in the information structure of a computer system “, Cybernetics and Systems Analysis, Vol. 54б No. 2, pp. 142-150.
Task to prepare data for enhanced machine learning, available at:
https://docs.microsoft.com/en-us/azure/machine-learning/team-data-science-process/prepare-data.
Chandrashekar, G. and Sahin, F (2014), “Survey on feature selection methods”, Computers & Electrical Engineering, Vol. 40, No. 1, pp. 16-28.
Hu, Z. (2015), “Hybrid filter–wrapper feature selection for short-term load forecasting”, Engineering Applications of Artificial Intelligence, Vol. 40, No. 1, pp. 17-27.
Reif, M. and Shafait, F. (2014), “Efficient feature size reduction via predictive forward selection”, Pattern Recognition, Vol. 47, No. 4, pp. 1664-1673.
Osanaiye, O., Cai, H. Choo, K.-K. R., Dehghantanha, A., Xu, Z. and Dlodlo, M. (2016), “Ensemble-based multi-filter feature selection method for ddos detection in cloud computing”, EURASIP Journal on Wireless Communication sand Networking, No. 1, pp. 130-140.
Tibshirani, R.J. (2016), “Exact post-selection inference for sequential regression procedures”, Journal of the American Statisti-cal Association, Vol. 111, No. 514, pp. 600-620.
Dyakonov A.G. (2014), “Methods for solving classification problems with categorical features“, Applied Mathematics and Computer Science, Vol. 46, pp. 103-127.