BankSealer: A decision support system for online banking fraud analysis and investigation

Carminati, Michele; Caron, Roberto; Maggi, Federico; Epifani, Ilenia; Zanero, Stefano

doi:10.1016/j.cose.2015.04.002

Cited by 76 publications

(57 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…While the experiments presented in [8] showed that Banksealer is an effective approach in identifying frauds and seems to provide a meaningful support to the banking analysts in fraud investigation, one of its main limitations is the lack of a bi-directional communication channel between the unsupervised system and the banking analyst to collect and exploit analyst's feedback and knowledge to improve detection performance. Furthermore, Banksealer works with an empirical configurations of weights, manually set by analysts.…”

Section: Research Goalmentioning

confidence: 99%

“…Therefore, as already successfully done in [7,8], we inject fraud scenarios that replicate the typical attacks performed against online banking users. We consider two types of fraudulent attacks, reconstructed using domain expert and analyst advice:…”

Section: Synthetic Fraud Scenariosmentioning

confidence: 99%

“…While the experiments presented in [8] showed that Banksealer is an effective approach in identifying frauds and seems to provide a meaningful support to the banking analysts in fraud investigation, one of its main limitations is the inability to collect and exploit banking analyst's feedback. It also depended on a complex set of parameters that had to be tuned by hand before operations.…”

Section: Introductionmentioning

confidence: 99%

“…Thanks to the collaboration with a major banking group, we were able to develop Banksealer [7,8], a novel, unsupervised fraud analysis system that automatically ranks frauds and anomalies in banking transactions. While the experiments presented in [8] showed that Banksealer is an effective approach in identifying frauds and seems to provide a meaningful support to the banking analysts in fraud investigation, one of its main limitations is the inability to collect and exploit banking analyst's feedback.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

A Supervised Auto-Tuning Approach for a Banking Fraud Detection System

Carminati

Valentini

Zanero

2017

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

In this paper, we propose an extension to Banksealer, one of the most recent and effective banking fraud detection systems. In particular, until now Banksealer was unable to exploit analyst feedback to self-tune and improve its performance. It also depended on a complex set of parameters that had to be tuned by hand before operations. To overcome both these limitations, we propose a supervised evolutionary wrapper approach, that considers analyst's feedbacks on fraudulent transactions to automatically tune feature weighting and improve Banksealer's detection performance. We do so by means of a multiobjective genetic algorithm. We deployed our solution in a real-world setting of a large national banking group and conducted an in-depth experimental evaluation. We show that the proposed system was able to detect sophisticated frauds, improving Banksealer's performance of up to 35% in some cases.

show abstract

Section: Research Goalmentioning

confidence: 99%

Section: Synthetic Fraud Scenariosmentioning

confidence: 99%

“…While the experiments presented in [8] showed that Banksealer is an effective approach in identifying frauds and seems to provide a meaningful support to the banking analysts in fraud investigation, one of its main limitations is the inability to collect and exploit banking analyst's feedback. It also depended on a complex set of parameters that had to be tuned by hand before operations.…”

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Supervised Auto-Tuning Approach for a Banking Fraud Detection System

Carminati

Valentini

Zanero

2017

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…Carminati [5] suggested the BANKSEALER system which is a decision support system for analyzing online banking fraud. During the training phase of this system, easy models were developed to understand spending habits of customers based on past transactions.…”

Section: Introductionmentioning

confidence: 99%

Detection of Violations in Credit Cards of Banks and Financial Institutions based on Artificial Neural Network and Metaheuristic Optimization Algorithm

Monirzadeh¹,

Habibzadeh²,

Farajian³

2018

ijacsa

View full text Add to dashboard Cite

Abstract-Due to popularity of the World Wide Web and ecommerce, electronic communications between people and different organizations through virtual world of the Internet have provided a good basis for commercial and economic relations. These developments, although occurring for less than a century, electronic communications have always been subject to interference, cheating, fraud, and other acts of sabotage. Along with this increase in trading volume, there is a huge increase in the number of online fraud which results in billions of dollars of losses annually worldwide; this has a direct effect on customer service of banking systems, particularly electronic banking systems, and survival as a reliable financial service provider. Therefore, attention to fraud detection techniques is essential to prevent fraudulent acts and is the motive for many scientific researches. For this reason, business intelligence is used to identify financial violations in various economic, banking and other fields. Here, the focus is on algorithms and methods presented in data mining to deal with fraud by using neural networks. The main objective is to improve these methods or present new algorithms by studying the behavioral patterns of customers and the combined use of genetic algorithm to improve the performance of neural network and find the appropriate models for better decision making by implementing and testing the performance of the suggested algorithms. The results show that more strength was given to neural network by using genetic algorithm. In fact, genetic algorithm can raise our ability to control the training process. Moreover, it was concluded that criteria such as age, gender, marital status were not effective on detection; in fact, the most important effective criteria are information related to transaction.

show abstract