Scholarlink

<p>Currently, credit card frauds are rising significantly due to the advancement and growth of internet technologies. Hence, most of financial institutions face complexities to predict transactions that are made with credit cards. Recent statistical reports indicate that millions of dollars are lost every year due to the fraudulent activities of credit card scams. Fraudsters are constantly looking for new laws and techniques to commit crimes. Technology for fraud protection has therefore become crucial in order to prevent losses for banks as well as other financial organizations. In earlier investigations, a variety of statistical techniques and machine learning models were used to build a framework for detecting credit card fraud. However, it has problems with inaccurate predictions, misleading predictions, high time requirements, low accuracy, and complex system architecture. The original contribution of this work is to develop a unique and smart security framework for identifying credit card frauds from the financial sectors. It helps to ensure the secured transactions in the financial institutions with low computational cost and system complexity. For accomplishing this objective, the sophisticated iterative weighted feature selection (IWFS) technique integrated with spiking imperialist competitive recurrent neural network (SICRNN) classification model is developed in this study. In the proposed framework, the customer transaction data obtained from the UCI repositories have been used to test the proposed system. Once the customer transaction data has been gathered, procedures such as data cleansing, mapping, transformation, and normalization are carried out to provide an attribute-balanced dataset. Then, the dimensionality of input transaction data is reduced using an intelligent IWFS mechanism by picking the most relevant features. Consequently, the SICRNN mechanism is implemented to take appropriate and correct decisions for identifying fraudulent credit card transactions. During classification, an Imperialist Competitive Optimization Algorithm is deployed to reduce the hidden states in accordance with the best solution. In addition, the performance of the proposed IWFS-SICRNN model is independently evaluated and compared using various parameters.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Sophisticated Iterative Weighted Feature Selection (IWFS) Based Spiking Imperialist Competitive Recurrent Neural Network (SICRNN) Classification Model for Credit Card Fraud Detection

  • 期刊论文

  • 发表日期：

    2025年2月26日

查看全文
  • S. Sobana,
  • V. Diana Earshia,
  • R. Suganthi,
  • K. Ayyappa Swamy

摘要

Currently, credit card frauds are rising significantly due to the advancement and growth of internet technologies. Hence, most of financial institutions face complexities to predict transactions that are made with credit cards. Recent statistical reports indicate that millions of dollars are lost every year due to the fraudulent activities of credit card scams. Fraudsters are constantly looking for new laws and techniques to commit crimes. Technology for fraud protection has therefore become crucial in order to prevent losses for banks as well as other financial organizations. In earlier investigations, a variety of statistical techniques and machine learning models were used to build a framework for detecting credit card fraud. However, it has problems with inaccurate predictions, misleading predictions, high time requirements, low accuracy, and complex system architecture. The original contribution of this work is to develop a unique and smart security framework for identifying credit card frauds from the financial sectors. It helps to ensure the secured transactions in the financial institutions with low computational cost and system complexity. For accomplishing this objective, the sophisticated iterative weighted feature selection (IWFS) technique integrated with spiking imperialist competitive recurrent neural network (SICRNN) classification model is developed in this study. In the proposed framework, the customer transaction data obtained from the UCI repositories have been used to test the proposed system. Once the customer transaction data has been gathered, procedures such as data cleansing, mapping, transformation, and normalization are carried out to provide an attribute-balanced dataset. Then, the dimensionality of input transaction data is reduced using an intelligent IWFS mechanism by picking the most relevant features. Consequently, the SICRNN mechanism is implemented to take appropriate and correct decisions for identifying fraudulent credit card transactions. During classification, an Imperialist Competitive Optimization Algorithm is deployed to reduce the hidden states in accordance with the best solution. In addition, the performance of the proposed IWFS-SICRNN model is independently evaluated and compared using various parameters.