Optimizing Credit Card Fraud Detection with Multi-Armed Bandit Algorithms

Rongjun Gao

2024

Abstract

In today's world, the importance of credit card fraud detection cannot be overstated, as it is crucial for the security of financial transactions. To optimize cost-efficiency, automated algorithms have been developed to pinpoint the transactions that are most likely to be fraudulent. Despite their potential, multi-armed bandit (MAB) algorithms have not been widely adopted in fraud detection. This paper introduces two models that apply the Upper Confidence Bound 1 and Thompson sampling algorithms to the task of fraud detection, categorizing transactions into 52 segments based on the amount and type. The performance of these algorithms is evaluated against several metrics, including cumulative regret, the reward generated, the ratio of optimal arm selection, and overall efficiency. The findings suggest that the Thompson sampling algorithm surpasses the UCB1 in performance, achieving lower standard errors and computational complexity. It proves to be more effective in swiftly and accurately identifying the most suspicious transactions, thus pinpointing the optimal choice with greater speed.

Paper Citation

in Harvard Style

Gao R. (2024). Optimizing Credit Card Fraud Detection with Multi-Armed Bandit Algorithms. In Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI; ISBN 978-989-758-713-9, SciTePress, pages 474-478. DOI: 10.5220/0012956000004508

in Bibtex Style

@conference{emiti24,
author={Rongjun Gao},
title={Optimizing Credit Card Fraud Detection with Multi-Armed Bandit Algorithms},
booktitle={Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI},
year={2024},
pages={474-478},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012956000004508},
isbn={978-989-758-713-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI
TI - Optimizing Credit Card Fraud Detection with Multi-Armed Bandit Algorithms
SN - 978-989-758-713-9
AU - Gao R.
PY - 2024
SP - 474
EP - 478
DO - 10.5220/0012956000004508
PB - SciTePress