Is Machine Learning Changing the Face of Fraud Prevention?
News
October 20, 2016
Ciaran Carr, MRC European Program Manager
Machine learning has been occupying the thoughts of fraud practitioners in recent years, with many questioning whether this revolutionary concept can replace traditional rule-based fraud detection systems all together. The conference circuit has seen a range of panel discussions, workshops and briefings dedicated to sharing knowledge and best practice techniques; the Merchant Risk Council's (MRC) recent Fall Platinum events in Vienna and Atlanta included.
Today, merchants need fraud detection tools that work at the same pace as modern business. Machine learning uncovers patterns in huge amounts of historical and live data to identify fraudulent transactions. The algorithms which underpin machine learning make predictions in real-time about the integrity of a transaction based on data from genuine and fraudulent transactions. Although primarily discussed in the context of fraud prevention, machine learning can also play a role as a transaction approval tool. By evaluating patterns in good transactions, it can draw distinctions between good and fraudulent transactions and separate them into two distinct groups.
As merchants strive to provide customers with a safe payment experience, among other uses, we see machine learning being deployed as part of automated fraud screening systems, identifying high-risk transactions, profiles and accounts, detecting and analysing changes in user behaviour and combatting account takeover.
Although it is at the forefront of our thinking today, the methodologies that underpin machine learning are nothing new and certainly not exclusive to fraud detection. In fact, frequently referenced examples of how algorithms influence our daily lives include movies and box sets recommended to us by the likes of Netflix, and Spotify's music suggestions made based on our listening habits and taste.
Weighing up the pros and cons of machine learning
Machine learning is often scrutinised through a lens of what traditional rule-based systems can or cannot do well. Using rule-based systems as a basis, let's look at some key strengths of machine learning...
Read the entire article in The Paypers' 2016 Web Fraud Market Guide published this December.
About Ciaran
Ciaran Carr, MRC European Program Manager, develops existing and new programs, education and resources to support MRC members. Ciaran has a background in Human Resource Management and Industrial Relations, and recently completed an MSc in Project & Program Management.
Machine learning has been occupying the thoughts of fraud practitioners in recent years, with many questioning whether this revolutionary concept can replace traditional rule-based fraud detection systems all together. The conference circuit has seen a range of panel discussions, workshops and briefings dedicated to sharing knowledge and best practice techniques; the Merchant Risk Council's (MRC) recent Fall Platinum events in Vienna and Atlanta included.
Today, merchants need fraud detection tools that work at the same pace as modern business. Machine learning uncovers patterns in huge amounts of historical and live data to identify fraudulent transactions. The algorithms which underpin machine learning make predictions in real-time about the integrity of a transaction based on data from genuine and fraudulent transactions. Although primarily discussed in the context of fraud prevention, machine learning can also play a role as a transaction approval tool. By evaluating patterns in good transactions, it can draw distinctions between good and fraudulent transactions and separate them into two distinct groups.
As merchants strive to provide customers with a safe payment experience, among other uses, we see machine learning being deployed as part of automated fraud screening systems, identifying high-risk transactions, profiles and accounts, detecting and analysing changes in user behaviour and combatting account takeover.
Although it is at the forefront of our thinking today, the methodologies that underpin machine learning are nothing new and certainly not exclusive to fraud detection. In fact, frequently referenced examples of how algorithms influence our daily lives include movies and box sets recommended to us by the likes of Netflix, and Spotify's music suggestions made based on our listening habits and taste.
Weighing up the pros and cons of machine learning
Machine learning is often scrutinised through a lens of what traditional rule-based systems can or cannot do well. Using rule-based systems as a basis, let's look at some key strengths of machine learning...
Read the entire article in The Paypers' 2016 Web Fraud Market Guide published this December.
About Ciaran
Ciaran Carr, MRC European Program Manager, develops existing and new programs, education and resources to support MRC members. Ciaran has a background in Human Resource Management and Industrial Relations, and recently completed an MSc in Project & Program Management.
Share: