Whitebox Machine Learning
Updated on 06.07.22
5 minutes to read
In SEON, whitebox machine learning algorithms are trained on your historical data to identify patterns and behaviors. You can then implement rules to block certain user actions, such as suspicious logins, identity theft, or fraudulent transactions.
SEON's machine learning tools
Glance at our different machine learning tools to understand how they differ.
|Automatic retraining several times a day||Yes||Yes|
|Uncovers complex fraud patterns||Yes||Yes|
|Automatic rule creation||Yes||No|
|Can be used in rules||No||Yes|
|Can effect fraud score||No||Yes|
|Can change transaction state||Yes||Yes|
|Fraud probability scoring||No||Yes|
|Keeps you in control||Yes||Yes|
|Fully automatable as needed||Yes||Yes|
|Available from Day 1||No*||Yes|
* Whitebox machine learning requires account-specific training data to begin recommending rules: at least 1000 transactions with 100 in the DECLINE and 100 in the APPROVE state.
Using whitebox machine learning
The algorithm retrains itself numerous times a day and creates human-readable rule suggestions with specific accuracy percentages. Our dedicated data scientists can also help you with resources and reporting.
Visit the Machine Learning tab of the Scoring Engine to review the rules created by the whitebox algorithm. Machine learning rules are divided into two rule categories by default: Complex rules and Heuristic rules.
The algorithm will automatically calculate the accuracy of all rules it creates based on past transactions. Accuracy scores compare the number of declined and accepted transactions the rule would affect. For example, a rule that would affect ten past transactions is 90% accurate if nine of these transactions are in the DECLINE state and only one in APPROVE.
You can choose to enable these rules above a set accuracy threshold automatically or turn them on or off manually at any time by using the toggle on the rule list.
Complex rules are based on surprising connections between data points. These are flagged in past and future transactions. Complex rules contain several parameters and data points.
These rules are designed to decline transactions from fraudulent accounts after the second offense. Heuristic rules target a single identified parameter (e.g., IP=X).
Review and enable rules
- Head to the Scoring Engine.
- Open the Machine Learning tab.
- Here, you can review the rules created by the algorithm. Click a rule to check details.
- Use the toggle on the left to enable and disable rules.
Machine Learning rule details
When you click on a machine learning rule, the modal includes advanced options.
Click Test rule on existing data to recalculate the accuracy of a rule on fresh data. The Filter transactions button will take you to a list of all past transactions the rule would affect. Choose Add to custom rules if you'd like to tweak the parameters of the rule before turning it on.
The Rule Details modal includes advanced options to help you use whitebox rules.
Whitebox machine learning is enabled on your account by default. However, we won't turn rules on automatically unless you change your settings.
The Machine Learning tab of the Settings page houses all settings related to your whitebox and blackbox models. Find settings related to the whitebox model divided into Complex Rule Settings and Heuristic Rule Settings.
Complex rule settings
Using these settings, you can tell SEON to enable new Complex rules in your account automatically. All you need to do is set an accuracy threshold, and we'll do the rest.
The algorithm will also maintain these rules in the long run – if a rule's accuracy falls below the threshold you set here, we'll turn it off automatically.
Heuristic rule settings
These settings will tell SEON to automatically enable any heuristic rules that meet the set criteria in your account.
Click the toggle and specify the transaction data fields that should be included in rules that turn on automatically. Finally, set an accuracy threshold, and we'll do the rest.
Read more about how SEON harnesses the power, speed, and accuracy of machine learning and what you can do to get the best results.