FraudIA™
Making banking transactions
more secure
Quick links:
FraudIA™ was originally developed by Transactis a subsidiary of Société Générale and la Banque Postale specializing in payment solutoins.™ This payment card detection solution™ is designed to be used in addition to the expert rules systems in fraud prevention software.
Plus points
FraudIA™
of fraudulent transactions
other software misses*
transactions per day scanned
for four banks
maximum response time
for faster transactions
Easy to use
Interfaces with your current software
Proven
Tested on volume metrics in real time
Predictive
Using artificial intelligence and continuous learning
Secure
Validated on customer intrusion tests
*Estimations based on simulations on 2017-2019 customer data. The percentage of fraudulent transactions detected ranges from 5% to 35% depending on the type of transaction (e-commerce, withdrawals, payments) and the bank. The percentage presented here is the percentage of fraudulent transactions reported as detected by the FraudIA™ algorithm, with a degree of accuracy of 30%. A fraudulent transaction reported means one that is reported by the bank’s customer; in other words, a fraudulent transaction that was not detected by any of the bank’s other tools.
Actual performance may vary, and is highly dependent on the configuration of the other fraud detection tools in place and changes in the behaviors of perpetrators of fraud and banking customers from the time the model is developed to when it is implemented.
AI helps reduce payment fraud on e-commerce websites
FraudIA™™ helps reduce the number of fraudulent payment cards and transactions for more secure e-commerce, withdrawals, and payments.
Banking transactions depend on fast processing. FraudIA™™ responds to authorization requests in just 50 milliseconds.
FraudIA™ very rapidly processes high volumes of data optimized to support hundreds of transactions per second:
- The transaction history associated with a payment card is analyzed
- The indicators of fraud are calculated
- The prediction is modelled
FraudIA™
FraudIA™™ uses artificial intelligence to identify the likelihood that a transaction is fraudulent in real time based on:
- Information about the transaction
- The transaction history associated with the bank card being used
- The transaction histories of all payment cards
- The transaction histories of payment cards cancelled due to fraud
- The fraudulent transaction history
FraudIA™™ uses a dynamic self-learning algorithm to build detection models. It analyzes fraudulent transaction and cancelled payment card data to identify new fraud scenarios.
FraudIA™ was created by a team of data scientists, developers, and architects™ to come up with the optimal detection model based customers’ actual data.
A FraudIA™ deployment
at Transactis
Transactis, co-owned by la Banque Postale and Société Générale, handling payment and SEPA and international transfer and direct debit systems and services for both banks.
Challenge
Transactis, one of France’s leading payment providers, handles huge volumes of data. Every day, more than 3 million payment cards are used to carry out some 6 million transactions. Every year, nearly 1 terabyte of data is generated by the 15 million payment cards in circulation managed by Transactis.
The project
Probayes used Big Data technologies and NoSQL to tackle the huge data volumes involced in this project. Apache Kafka processes the data flows between the machines in the cluster. Couchbase and RocksDB in-memory databases were deployed to ensure extremely fast data logging response times. Kafka was also used for node-to-node messaging, with real-time rating, model training, and response.
Containers and an orchestrator were implemented to isolate the processes and ensure high uptime and scalability.
Results
The solution has been deployed and performance data will soon be available.
“Probayes offers an AI-based, real-time fraud prevention solution to the banking and insurance industry.
FraudIA™ can handle more than 200 transactions per second with a maximum response time of 50 milliseconds per transaction.
The solution, designed to be deployed in addition to the customer’s current fraud prevention software, detects fraudulent withdrawal and e-commerce transaction patterns the software misses. These tools, when used together, help reduce fraud”.
Vincent Maigron
Probayes solutions for the financial services industry
FraudIA™
Detect credit card fraud
Churn
Detecting customers that are likely to leave
Chatbot
Giving customers and employees a conversation agent
Copyright 2019 – Probayes – 53 avenue Jean Kuntzmann – 38330 Montbonnot – France – Tel: +33 (0)4 76 42 64 13
Mentions légales et conditions générales d’utilisations – Politique de protection des données du site internet de Probayes – Politique Cookies