COMPARISON OF THE EFFECTIVENESS OF ANOMALY DETECTION BY MACHINE LEARNING ALGORITHMS WITHOUT A TEACHER

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