Quarterly journal published in SPbPU
and edited by prof. Dmitry Zegzhda
Peter the Great St. Petersburg Polytechnic University
Institute of computer sciences and technologies
information security of computer systems
Information Security Problems. Computer Systems
Published since 1999.
ISSN 2071-8217
COMPARISON OF THE EFFECTIVENESS OF ANOMALY DETECTION BY MACHINE LEARNING ALGORITHMS WITHOUT A TEACHER

N. V. Gololobov, E.Y. Pavlenko
Peter the Great St. Petersburg Polytechnic University (SPbPU)

Annotation: The paper proposes the use of recurrent neural networks with the LSTM architecture for solving problems related to the detection of anomalous instances in data sets and compares the effectiveness of the proposed method with the traditional technique – the support vector machine for one class. During the study, an experiment was conducted and criteria for the effectiveness of implementations were formulated. The results obtained in this way made it possible to draw appropriate conclusions about the applicability of recurrent neural networks in the tasks of detecting anomalous instances and put forward proposals for the further development of this direction.
Keywords: anomaly detection, machine learning, support vector method, recurrent neural networks, LSTM, learning without a teacher, recurrent neural networks
Pages 135-147