Soft Segmentation for False Positive Reduction in Rule-Based Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Rule-based systems for detecting financial crimes, such as money laundering and fraud, often produce high volumes of false positives, requiring significant human resources to manage, while maintaining true positive detection, and are restrictive in not allowing new true positives in constrained systems like anti-money laundering.
Innovation Solution
The system employs soft-segment based rules optimization using a topic model to determine semantic structures and entity behavior archetypes, allowing real-time re-assignment of accounts and adjusting rule parameters to reduce false positives without losing true positives, by applying Bayesian inference and latent archetype updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If rule-based systems are used to detect financial crimes, then true positive detection is maintained, but false positive volume increases significantly
Solution Approach 1:
The patent segments the transaction monitoring system into multiple rule sets with different specificity levels. High-specificity rules are applied first to catch obvious crimes, followed by lower-specificity rules that capture broader patterns. This segmentation allows the system to maintain high true positive detection while reducing false positives by organizing rules hierarchically rather than applying them uniformly.
Solution Approach 2:
The patent applies different rule parameters and thresholds locally to different account types, transaction categories, and risk profiles. Instead of using uniform rules across all transactions, the system tailors rule sensitivity and parameters to specific contexts, which reduces false positives for low-risk transactions while maintaining detection capability for high-risk scenarios.
2Quantity of substance
If analytic models are used to reduce false positives, then false positive volume decreases, but true positive detection may be lost
Solution Approach 1:
The patent implements feedback mechanisms where the system continuously monitors rule performance metrics including false positive rates and true positive detection rates. Based on this feedback, the system automatically adjusts rule parameters, thresholds, and activation conditions to optimize the balance between reducing false positives and maintaining true positive detection. This closed-loop control ensures that false positive reduction does not come at the cost of missing crimes.
Solution Approach 2:
The patent employs dynamic rule parameters that adapt to changing transaction patterns, account behaviors, and emerging crime techniques. Rules can be activated, deactivated, or have their thresholds adjusted in real-time based on system performance and external factors, allowing the system to maintain optimal detection accuracy while minimizing false positives across different operational conditions.
3Reliability
If constrained systems are used in anti-money laundering, then regulatory compliance is maintained, but system flexibility and optimization are restricted
Solution Approach 1:
The patent structures the detection system with nested rule sets where core regulatory-compliant rules form the inner layer, ensuring mandatory detection requirements are met. Outer layers contain additional analytical rules and models that provide enhanced detection capabilities while operating within the framework established by the inner compliant rules. This nesting allows the system to maintain regulatory compliance as the foundation while adding flexible optimization capabilities on top.
Data Source
AI summary
A system and method includes soft-segment based rules optimization that can mitigate the overall false positives while maintaining 100% true positive detection. The soft clustering allows real-time re-assignment of an account to a dominate archetype behavior, as well as rule optimization based on a logical order with more relaxation on thresholds for the most inefficient rules is performed within each archetype. The rule optimization provides false positive reduction compared to a baseline rule system. The method can be used to reduce false positives for any rule-based detection system in which the same true positive detection is required.