Next generation fraud prevention platforms leverage ML to secure payments

As fraudsters follow the growth of the cashless economy online, anti-fraud companies are building powerful tools and techniques that mine various data for fraudulent behaviour patterns.

Fraudulent attacks are getting to be more sophisticated and inventive. Once a new solution against fraud is developed, fraudsters immediately find a new loophole. And it seems that the risk professionals are always a step behind.
Machine learning can be used to help solve this problem, but at the moment it is impossible to completely abandon human intervention.

Rule-based and machine learning approaches complement each other because machines can analyse a larger volume of characteristics, based on the context, while risk analysts can create models that are easily understood by humans, unlike the machine-learning approach alone. Each industry has its own unique set of features and each fraud prevention system aims to adapt them to avoid false positives (good customers identified as fraudsters) and false negatives (fraudsters identified as good customers). Moreover, the risk system needs to periodically be examined by risk managers and afterwards tuned, for example, if online merchants sell new products or make frequent changes to their billing logic.

The power of supervised and unsupervised machine learning

A machine learning solution needs access to a big store of historical data to train its models and increase the probability that it will uncover patterns of new suspicious activity. The more data, the better the system becomes at detecting and preventing fraud.

The machine learning process contributes to the learning of nonlinear combinations of latent characteristics and their combinations that lead to predictiveness enhancement.

There are two approaches that are used in machine learning: supervised and unsupervised learning. The first approach is the most common and widespread.

With the supervised approach, in the beginning, a risk analyst creates a machine learning model based on historical data. Then, with new transaction data, the algorithm creates potentially right baskets: fraud and not fraud. After that, the system collects external signals such as fraud alerts, chargebacks, complaints etc. Based on that information, the algorithm starts looking for new unrecorded dependencies. Finally, the model starts retraining. Consequently, all the risk analysts are one step behind the game, thus, the cycle continues and with time new techniques emerge.

Unsupervised learning is regarded as an alternative to supervised learning. These algorithms infer patterns from a dataset without reference to known or labelled outcomes. Unsupervised learning allows risk analysts to approach problems with no exact idea about what the result will look like. One can derive structure from data where they don’t necessarily know the effect of the variables. With unsupervised learning, there is no feedback based on the prediction results. But it can divide data on the basis of anomalous behaviour and then risk analysts can apply well-known supervised approaches to this data.

Therefore, unsupervised machine learning is more applicable to real-world problems and can help to solve them when risk managers are constantly one step behind the fraudsters.

Why use machine learning in payment fraud prevention?

When it comes to detecting and fighting online payment fraud, several advantages become evident:

it facilitates real-time decision-making and improves the experience for customers;

it improves accuracy of classification;

it helps detect new fraudulent behaviour;

it provides a more rapid response to real-world changes.

What can the best fraud prevention solutions do

The most advanced fraud prevention services use both rule-based and machine learning approaches, including unsupervised techniques, with an industry focus and an adaptation for the business’ individual characteristics and customer needs. The result is a solution that makes more accurate decisions for each industry and every customer. One of the companies working in this space is called Covery. Risk analysts can customise any combination of data patterns we call ‘features’ that can be applied to a specific business needs. Covery can also accept any non-payment data in any user action to supplement the profile with missing details to analyse by using both rule-based and machine learning models for more precise decisions.

So what is Covery?

Covery is a global risk management platform helping online companies solve fraud and minimise risk. The company focuses on the versatility of the product and its adaptability to each type of business, based on the individual characteristics and customer needs using both rule-based and machine learning approaches. Covery works with high-risk as well as with low-risk industries to find the right solution for every customer.

What Covery offers to help with fraud prevention:

wider coverage of user actions for analysis

flexible customisation of data patterns

usage of any additional data for analysis

rule-based and machine learning approaches

functionality to work with loyal users to increase revenue

custom machine learning models creation

custom functionality upon request.

Conclusions

Fraudsters are always developing new tricks and risk managers don’t always have the time to adapt to new changes. Machine learning has long been expected to help solve the problem of preventing fraud, but the majority of solutions are still on the path of development. So Covery’s main goal is to solve the problem when risk managers are constantly one step behind the fraudster.

About Pavel Gnatenko

Pavel has a master’s degree in intellectual systems for decision-making. He is a risk management expert with more than seven years of experience in the fintech industry. Currently, Pavel is focused on developing Covery - next generation of risk management platforms.

About Covery

Covery is a global risk management platform helping online companies solve fraud and minimize risk. We focus on the universality of our product and its adaptation to any type of business, based on the individual characteristics and customer needs using both rule-based and machine learning approaches. Also, Covery team publishes educational comic books helping risk managers to find new methods of fraud detection and risk assessment.