Use of Machine Learning to Detect Credit Card Fraud

EasyChair Preprint 8307

Abstract

For business owners, payment card issuers and transaction service providers, payment card fraud is a major problem, leading to significant and growing financial losses every year. As we all know, detecting fraudulent systems in payment card or plastic money dealings and transactions is a very complicated task. As the amount of data produced by payment card or plastic money dealings and transactions continues to increase, it is not possible for human analysts to identify fraud patterns in dealings and transaction records, which are usually characterized by high sample numbers, multi-dimensional and online updates. In the past decade, the development of methods for detecting payment card fraud has increasingly focused on machine learning (ML) methods, which automate the process of detecting large-scale fraud programs.

Integrating the LD method into the credit card fraud detection system greatly improves their ability to detect fraud more effectively and help payment intermediaries detect illegal transactions. In 2016, the number of fraud cases started to decline-a counter-trend related to the increasing adoption of machine learning solutions. Now, the introduction of machine learning-based fraud detection systems not only helps save money, but also becomes an obligation for institutions and companies to gain customer trust.

In this emerging DL field of DL card fraud detection, a well-known problem that persists is that most published studies on this topic lack reproducibility. On the one hand, due to confidentiality reasons, no data on payment card transactions cannot be released. On the other hand, the authors did not go to great lengths to provide their code and make the results reproducible.

Keyphrases: Max Pooling, Transfer Learning, convolution, machine learning

@booklet{EasyChair:8307,
  author    = {Sachin Agwan and Nitin Patil},
  title     = {Use of Machine Learning to Detect Credit Card Fraud},
  howpublished = {EasyChair Preprint 8307},
  year      = {EasyChair, 2022}}