Neural Network Exclude List for False Decline Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for transaction fraud prevention often result in false declines, where legitimate transactions are incorrectly flagged as fraudulent, leading to significant losses for issuers and merchants due to the inability to update real-time decisioning rules effectively and efficiently.
Innovation Solution
A computer-implemented method using a neural network trained on prior transaction data to generate an exclude account list, which identifies transactions that are likely to be falsely declined, allowing subsequent transactions to be authorized without applying real-time decisioning rules, thus reducing false declines and optimizing the objective function based on probabilities and amounts of transactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If real-time decisioning rules are applied to all transactions to prevent fraud, then fraud detection capability is improved, but false decline rate increases
Solution Approach 1:
The patent segments transactions into different groups based on fraud risk characteristics. High-risk transactions are subjected to strict real-time decisioning rules, while low-risk transactions are routed to an exclude account list that allows faster processing. This segmentation enables differentiated fraud prevention strategies that reduce false declines for legitimate transactions while maintaining strong security for suspicious ones.
Solution Approach 2:
The system performs preliminary analysis of transaction patterns and proactively creates exclude account lists before fraudulent transactions occur. By analyzing historical data and transaction behaviors in advance, the system identifies accounts that are likely to be legitimate and pre-approves them, preventing false declines before they happen.
2Adaptability or versatility
If real-time decisioning rules are updated frequently to adapt to new fraud patterns, then fraud prevention effectiveness is improved, but system complexity increases
Solution Approach 1:
The system automatically updates the exclude account list based on incoming transaction data without requiring manual intervention. The neural network continuously learns from new transactions and self-adjusts the exclude account list, enabling the system to adapt to new fraud patterns autonomously while reducing operational complexity.
Solution Approach 2:
The exclude account list is dynamically updated in real-time as new transaction data arrives. The system transitions from static fraud prevention rules to dynamic, adaptive lists that automatically adjust based on current transaction patterns, enabling continuous improvement without manual rule updates.
3Manufacturing precision
If neural network processing is applied to all transactions to generate exclude account lists, then false decline mitigation is improved, but processing time increases
Solution Approach 1:
The system applies neural network processing selectively to a subset of transactions that benefit most from exclude account list generation. Rather than processing every transaction through the full neural network pipeline, the system identifies high-value transactions or those with specific characteristics that warrant exclude list creation, reducing overall processing time while maintaining effective false decline mitigation where it matters most.
Data Source
AI summary
A method, system, and computer program product for false decline mitigation. The method includes obtaining an objective function associated with an issuer system; training a neural network, based on prior transaction data associated with one or more prior transactions, to optimize the objective function; providing the trained neural network; receiving transaction data generated, based on one or more case creation (CC) rules, during processing of a transaction associated with an account identifier; processing, using the trained neural network, the transaction data to generate an exclude account list including the account identifier; receiving subsequent transaction data associated with a subsequent transaction associated with the account identifier; and authorizing, based on the exclude account list and the account identifier, the subsequent transaction associated with the account identifier without applying one or more real-time decisioning (RTD) rules to the subsequent transaction.