Itsy Bitsy SpiderNet: Fully Connected Residual Network for Fraud Detection

TL;DR

This paper introduces SpiderNet, a fully connected residual neural network tailored for fraud detection, incorporating novel feature engineering methods inspired by Benford's Law, and demonstrates its superior performance over traditional models.

Contribution

The paper presents SpiderNet architecture and new fraud feature tests, B-tests and W-tests, advancing neural network-based fraud detection methods.

Findings

SpiderNet outperforms Random Forest and other neural networks in fraud detection.

B-tests and W-tests significantly improve model accuracy.

Proposed methods are effective in identifying fraud anomalies.

Abstract

With the development of high technology, the scope of fraud is increasing, resulting in annual losses of billions of dollars worldwide. The preventive protection measures become obsolete and vulnerable over time, so effective detective tools are needed. In this paper, we propose a convolutional neural network architecture SpiderNet designed to solve fraud detection problems. We noticed that the principles of pooling and convolutional layers in neural networks are very similar to the way antifraud analysts work when conducting investigations. Moreover, the skip-connections used in neural networks make the usage of features of various power in antifraud models possible. Our experiments have shown that SpiderNet provides better quality compared to Random Forest and adapted for antifraud modeling problems 1D-CNN, 1D-DenseNet, F-DenseNet neural networks. We also propose new approaches for fraud feature engineering called B-tests and W-tests, which generalize the concepts of Benford's Law for fraud anomalies detection. Our results showed that B-tests and W-tests give a significant increase to the quality of our anti-fraud models. The SpiderNet code is available at https://github.com/aasmirnova24/SpiderNet