Fraud detection

Client Needs & Objectives

Fighting financial fraud is a major banking issue :

  • Working with imbalanced data make fraud detection difficult
  • Strong increase of players in the financial service ecosystem impacting the BAUs processes (open banking, social media…)
  • Increase of selfcare operations directly taken by the customer
  • Strong regulatory requirements (KYC, MIFID…)

Our Approach

Use of AI/Machine learning to improve fraud detection through deep learning techniques, time series/sequence analysis and feature engineering techniques

Main Deliverables


Client Benefits & Main Results

AI / neural network techniques for time series helps detecting anomalies in a sequence / window

AI / Machine learning in fraud detection has several advantages:

- The ability to scale to millions or billions of Payment transactions

- The ability to use Graph / link networks to identify relationships

- Social media integration

- Performance – rapid response for acceptance / rejection of transaction

- Behavioral analysis/sentiment analysis

- Transparency – the ability to indicate the reason for rejection