Efficient loss updated XGBoost with deep emended genetic algorithm for detecting online fraudulent transactions

作     者：Lingeswari, R. Brindha, S. 

作者机构：Research Scholar Department of Computer Applications St.Peter’s Institute of Higher Education & ampResearch Tamilnadu Chennai India Department of Computer Science and Applications St.Peter’s Institute of Higher Education & ampResearch Tamilnadu Chennai India 

出 版 物：《Multimedia Tools and Applications》 (Multimedia Tools Appl)

年 卷 期：2024年第83卷第37期

页      面：84471-84494页

核心收录：

学科分类：0710[理学-生物学] 0831[工学-生物医学工程（可授工学、理学、医学学位）] 0810[工学-信息与通信工程] 08[工学] 0835[工学-软件工程] 0836[工学-生物工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：None 

主　　题：Genetic algorithms 

摘      要：In the fast-paced technological era, online financial transactions have gained widespread use as it offers significant merits to customers for easy transfer of money through smart phones. Nevertheless, fraudulent transactions put individual’s money into risk, for which, suitable approaches are required to detect such deceits. Concurrently, with the progress of ML (Machine Learning) approaches, existing works have bidden to identify the fraudulent and normal transactions. However, studies lacked in accordance with accuracy rate and only limited focus has been provided for detection of generalized fraudulent transactions. Considering this, the current study considers IoT fraud dataset and proposes DEGA (Deep Emended Genetic Algorithm) to attain better performance for detecting fraudulent and normal transactions. This model employs a competitive approach, integrating, new crossover and selection methods. This intend to improvise the ability of global search and partition the chromosomes into losers and winners. This ensures high quality parent for selection. Besides, a dynamic-mutation function is also proposed for enhancing the model’s searching ability. Subsequently, the study proposes EL-UXGB (Efficient Loss-Updated eXtreme Gradient Boosting) wherein dual sigmoid loss functions are proposed to resolve the imbalanced label cases. The overall performance of this study is assessed through analysis that confirms its effectiveness in detecting fraudulent transactions. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.