Financial Fraud Detection with Improved Neural Arithmetic Logic Units

Schlör, Daniel; Ring, Markus; Krause, Anna; Hotho, Andreas

doi:10.1007/978-3-030-66981-2_4

Cited by 6 publications

(4 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Indeed, the traditional SVM method is reportedly sensitive to outliers and noisy data (Shajalal et al, 2021). Table 7 also shows that deep neural networks performed well in previous studies (Schlör et al 2021;Xenopoulos, 2017). However, their performance is limited by the relatively low number of fraudulent transactions in the dataset.…”

Section: Comparison With State-of-the-art Methodsmentioning

confidence: 80%

“…To overcome this major limitation, class imbalance was first approached by using under-sampling methods and then machine learning methods were trained on the balanced dataset (Pambudi et al, 2019). Similarly, Xenopoulos (2017) used under-sampling to produce balanced bootstraps for ensemble learning, and Misra et al (2020) and Schlör et al (2021) applied it to generate balanced training data for deep learning-based detection models. The main drawback of the under-sampling approach is that potentially useful instances are often excluded from the training data, which can significantly degrade the detection accuracy.…”

Section: Systems -Literature Reviewmentioning

confidence: 99%

“…Previous studies have also considered the F1 measure (Pambudi et al, 2019;Schlör et al 2021), defined as the harmonic mean of precision and recall:…”

Section: Performance Evaluationmentioning

confidence: 99%

See 2 more Smart Citations

Fraud Detection in Mobile Payment Systems using an XGBoost-based Framework

2022

View full text Add to dashboard Cite

Mobile payment systems are becoming more popular due to the increase in the number of smartphones, which, in turn, attracts the interest of fraudsters. Extant research has therefore developed various fraud detection methods using supervised machine learning. However, sufficient labeled data are rarely available and their detection performance is negatively affected by the extreme class imbalance in financial fraud data. The purpose of this study is to propose an XGBoost-based fraud detection framework while considering the financial consequences of fraud detection systems. The framework was empirically validated on a large dataset of more than 6 million mobile transactions. To demonstrate the effectiveness of the proposed framework, we conducted a comparative evaluation of existing machine learning methods designed for modeling imbalanced data and outlier detection. The results suggest that in terms of standard classification measures, the proposed semi-supervised ensemble model integrating multiple unsupervised outlier detection algorithms and an XGBoost classifier achieves the best results, while the highest cost savings can be achieved by combining random under-sampling and XGBoost methods. This study has therefore financial implications for organizations to make appropriate decisions regarding the implementation of effective fraud detection systems.

show abstract

Section: Comparison With State-of-the-art Methodsmentioning

confidence: 80%

Section: Systems -Literature Reviewmentioning

confidence: 99%

See 1 more Smart Citation

Fraud Detection in Mobile Payment Systems using an XGBoost-based Framework

2022

View full text Add to dashboard Cite

show abstract

“…En otro estudio [9] los autores, implementaron una arquitectura de red neuronal que incorpora Unidades Lógicas Aritméticas Neurales Mejoradas recientemente propuestas. Se construyó un conjunto de datos sintéticos de referencia, el cual refleja el problema de capturar automáticamente tales relaciones matemáticas dentro de los datos; evaluaron dos conjuntos de datos de fraude financiero del mundo real y dos sintéticos para diferentes parámetros de red.…”

Section: Referente Teóricounclassified

Análisis del uso de técnicas supervisadas de aprendizaje automático y profundo en la detección de fraude financiero

Rodriguez-Tovar,

Gutiérrez-Portela,

Hernández-Aros

2023

View full text Add to dashboard Cite

En el mundo moderno se hace necesario el uso de técnicas, metodologías y acciones en busca de la integración de los diversos avances, herramientas y elementos vigentes para el trabajo conjunto en la solución a las problemáticas que afectan las finanzas de las organizaciones, puesto que ellas hacen que exista una dinámica empresarial, creando valor económico. Teniendo en cuenta lo anterior, en este estudio se analiza la prevención de fraudes empresariales, mediante el uso de técnicas de aprendizaje automático y profundo para generar prevención, tratamiento y resolución a los fraudes llevados a cabo en sistemas del orden financiero. A nivel metodológico, se obtuvo información en bases de datos a nivel documental, con fuentes fidedignas y estudios de caso, donde se prueba la efectividad en el uso de las técnicas anteriormente nombradas en la detección temprana del fraude empresarial. Los resultados obtenidos en los documentos consultados expresan que los algoritmos que presentan mayor efectividad en la prevención de estos fraudes son árbol de decisión, C5.0-SVM, Naïve Bayes y Random Forest, con porcentajes de: 92%, 83.15%, 80,4% y 76, 7% respectivamente. Frente al aprendizaje profundo, la literatura mostró que al hacer uso de unidades lógicas aritméticas neuronales y realizando la correcta clasificación de las neuronas iNALU y ReLU el porcentaje de efectividad incrementa en gran proporción. En la parte final de este documento se presentan y consolidan resultados y conclusiones, todo en el marco de la temática abordada, además la información recopilada en este documento está debidamente respaldada por los derechos de autor a quien corresponde.

show abstract