Application of Ensembles of Neural Networks Trained on Unbalanced Samples for Analyzing Statuses of IoT Devices.

Autor: Sukhoparov, M. E., Lebedev, I. S.
Zdroj: Automatic Control & Computer Sciences; Dec2021, Vol. 55 Issue 8, p1136-1141, 6p
Abstrakt: This article considers an approach to identifying abnormal situations in network segments of the Internet of Things by means of an ensemble of classifiers. Classification algorithms are adjusted for different kinds of events and anomalies with the help of training samples of various compositions. An ensemble of algorithms allows obtaining more accurate results by collective voting. An experiment using three neural networks with equal architectures is described. The estimation results are obtained both separately for each classifier and using the ensemble. [ABSTRACT FROM AUTHOR]
Databáze: Complementary Index
Externí odkaz: