Fraud analytics

Opening & welcome

Welcoming words

Joël Winteregg, CEO at NetGuardians

Vincent Peiris, Head of TIC at HEIG-VD

Journey to fighting fraud: INTERPOL’s perspective

Fraud is always evolving. In the face of increasing challenges and dealing with smart criminals, we need to continuously grow our team and be innovative in fighting it. In this talk, we will bring you through INTERPOL’s journey to combat fraud, specifically, the roles played by the Financial Crime Unit in connecting not just the respective Police agencies of its Member Countries, but also all related stakeholders.

Keynote speaker: Jung Kee You, Executive staff – Criminal Intelligence Officer at the Financial Crime Unit of INTERPOL

Fraud, a case of digital transformation

The issue of fraud is a use case that is particularly interesting from a machine learning standpoint. Modelizing this problem is rather complex from a technical perspective, as it is considered a rare occurrence; therefore, creating a robust model is quite difficult. Furthermore, it would necessitate close cooperation with the business in order to build a thorough relationship with them and to understand their needs. Working on this subject with them enables us to undergo a true cycle of digital transformation. The following presentation will explore this perspective.

Assia Garbinato, Expert in data transformation, Consultant

Kim Leng Chhun, Business Intelligence Designer & Data Scientist at La Vaudoise

Machine learning against fraud: lessons learned from a real case

Billions of dollars of loss are caused every year due to fraudulent credit card transactions. The design of efficient fraud detection algorithms is key for reducing these losses, and more and more algorithms rely on advanced machine learning techniques to assist fraud investigators. The design of fraud detection algorithms is however particularly challenging due to non-stationary distribution of the data, highly imbalanced classes distributions and continuous streams of transactions.  At the same time public data are scarcely available for confidentiality issues, leaving unanswered many questions about which is the best strategy to deal with them.

In this talk we will discuss a number of lessons learned during our long-standing collaboration with the R&D team of Worldline. In particular we will focus on best practices for the assessment of credit card fraud detection models and we will discuss the impact of data unbalancedness and non-stationarity on the resulting accuracy. More recent directions of research, including big data infrastructure, active and transfer learning, will be sketched as well.

 Gianluca Bontempi, Prof. at ULB & co-Head at MLG

Unusual bank transactions detector

How to detect unusual bank transactions? Based on a case study, this talk presents the process of building an unusual bank transactions detector using AI that leverages data acquisition and cleansing, features & algorithm selection and industrialization.

 Stephan Singh, Head of data analytics for regulatory & strategic projects at Pictet

Interpretable machine learning for credit card fraud detection

Credit card issuers put great effort into preventing fraud in order to minimise their own losses as well as the inconvenience caused to the affected cardholders. At Viseca Card Services we have been exploring various approaches regarding the integration of machine learning into our fraud prevention processes with a high degree of success. In this presentation we describe the use of machine learning to score Card-Not-Present transactions and new approaches to interpret the individual scores.

 Viton Vitanis, Head of Risk Analytics at Viseca Card Services

Data analytics on cyber crimes complaints registered at C3N, of Gendarmerie Nationale in France

Cyber risk cannot be avoided, whatever the implemented IT security measures. For improving society resilience to cyber-attacks, there is a strong need for understanding theses risks. A way towards this goal is the search for reliable data and their statistical analysis. It is what is proposed here, thanks to the partnership between the Center of Research in Econo-finance and Actuaria science on Risk (CREAR-ESSEC Business School) and the SCRC (service Central du Renseignement Criminel) of PJGN (Pôle Judiciaire de la Gendarmerie Nationale). We proceed to the statistical data exploration and analysis of the cyber crimes complaints by victims (individuals and companies) registered at the Gendarmerie National’s Cyber Crime Unit (C3N). Using an algorithmic method developed by Debbabi et al. (2017)2 for extremes modeling, we show the existence of heavy-tailed distributions, one of the main characteristics of cyber risk.

 Michel Dacorogna, Partner at Prime Re Solutions, and lecturer at ETHZ

Round table discussion

Fraudsters versus Data scientists, who is winning?

All speakers & Yves Genier, Business Journalist at La Liberté