Fraud Detection System Using Support Vector Machine, Feature Extraction and Cross Validation
Paper ID : 1132-IST
1hossein Gharaee *, 2babak rahmani
1ICt security dep, ITRC
In recent years, electronic payments have a rapid growth among internet activities; so it has attracted many customers due to its speed, efficiency, cost effectiveness and ease of access. Credit cards can be considered as one of the most widely used tools for electronic payments and transactions. This study aimed at identification and extraction features of fraudulent transaction in fraud detection followed by correct classification of them into two categories of legal and fraudulent features using support vector machine algorithm and cross-validation. The results of this method in comparison with others indicate improvement in fraud detection so that false negative errors have significantly reduced by 77% which leads to 88% reduction in costs and fraud detection rate increased by 11%.
Fraud detection, Credit cards, Support Vector Machine, Cross Validation, and Feature Extraction.