Detection of Fraudulent Behavior Using the Combined Algebraic and Machine Learning Approach

Abstract

The paper presents two approaches to detecting fraudulent behavior: the machine learning technique, which was developed to recognize fraud in the Internet purchasing system; the algebraic approach, which was developed for the detection of attacks in blockchain-based systems. The machine learning approach uses technique of translating heterogeneous data to homogenous data and builds the classification model based on fuzzy logic rules for fraud recognition. The algebraic approach uses behavior algebra methods and symbolic modeling techniques.