Answers are grounded in this paper and cite their sources.
The article describes an approach to detecting fraudulent calls using a neural network based on constructed statistical features of international telecommunication traffic calls. The research aimed to develop a monitoring module for an anti-fraud system, capable of countering a broad spectrum of attack types and solving the problem of binary classification of calls into fraudulent and legitimate. This module should use the entire available data volume from all existing sources for analysis, perform real-time statistical analysis of a set of calls to construct features, and classify calls based on these features. Realtime fraud detection will efficiently prevent various types of fraudulent activities before they can cause significant economic and reputational damage to the telecom operator. A literature review was conducted, the accumulated expertise in developing fraud prevention systems within the telecommunications sector was considered, the specific operations of a transit communication operator were taken into account. A feedforward neural network for the monitoring module of the anti-fraud system was developed and trained. The analysis of the obtained results and a comparison with the results of other authors' research were conducted, and the reasons for type I and type II errors were considered. Directions for further research were proposed to enhance the effectiveness of the monitoring module.
No takes yet. Share an insight, caveat, or question.
Santalov et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: