Dynamic Capacity Customization for Fraud Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional fraud detection systems are inefficient as they treat all transactions equally, leading to both missed fraudulent transactions and excessive false-positive identifications, resulting in increased costs and consumer frustration.
Innovation Solution
A multi-stage filtering process that dynamically adjusts server and application capacity based on transaction volume, utilizing preliminary filtration stages to quickly eliminate low-risk transactions and reserve resources for higher-risk ones, incorporating geo-positioning, velocity data, and customer history for more detailed analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional fraud detection systems analyze each transaction in the same manner, then fraud detection coverage is maintained, but false-positive identifications increase and legitimate transactions are declined
Solution Approach 1:
The patent segments the fraud detection process into multiple stages: a first stage that applies conventional uniform analysis to all transactions, and a second stage that applies enhanced analysis only to transactions flagged as potentially fraudulent. This segmentation allows the system to maintain broad fraud detection coverage while reducing false-positive identifications by avoiding unnecessary enhanced analysis on low-risk transactions.
Solution Approach 2:
The patent applies different levels of analysis quality to different transactions based on their risk profile. Transactions are locally differentiated into those requiring standard analysis and those requiring enhanced analysis, allowing the system to optimize detection precision for high-risk cases without compromising overall reliability.
2Reliability
If conventional fraud detection systems use strict filtering to eliminate fraudulent transactions, then fraud detection effectiveness improves, but false-positive identifications increase causing consumer frustration
Solution Approach 1:
The patent divides the transaction processing into two segments: standard processing for most transactions and enhanced processing only for flagged transactions. This reduces the overall impact of false-positive identifications by limiting strict filtering to only those transactions that truly require it, while maintaining fraud detection effectiveness through the enhanced second stage.
Solution Approach 2:
The patent applies partial enhanced analysis only to the subset of transactions that are flagged as potentially fraudulent, rather than applying strict filtering to all transactions. This partial action approach reduces false-positive identifications while maintaining sufficient fraud detection effectiveness through the targeted enhanced analysis.
3Measurement precision
If fraud detection systems process all transactions with full analysis, then detection accuracy is maintained, but computing and energy costs increase
Solution Approach 1:
The patent segments transaction processing into a low-cost first stage for all transactions and a high-cost second stage only for flagged transactions. This segmentation maintains detection accuracy for fraudulent transactions that require it while significantly reducing overall computing and energy costs by avoiding full analysis on the majority of low-risk transactions.
Solution Approach 2:
The patent applies partial enhanced analysis only to the necessary subset of transactions, avoiding excessive computing resources being wasted on transactions that do not require detailed analysis. This partial action approach optimizes the balance between detection accuracy and computational cost.
4Device complexity
If fraud detection systems apply uniform analysis to all transactions, then processing simplicity is maintained, but fraudulent transactions are missed and resources are wasted
Solution Approach 1:
The patent segments the processing approach into a simple first stage for all transactions and an enhanced second stage for flagged transactions. This segmentation maintains processing simplicity for the majority of transactions while improving fraud detection effectiveness through targeted enhanced analysis, resolving the contradiction between simplicity and effectiveness.
Solution Approach 2:
The patent performs preliminary analysis on all transactions to identify potential fraud cases before applying enhanced analysis. This preliminary action maintains processing simplicity for most transactions while preparing the system to apply more sophisticated analysis only where needed, thereby improving overall detection effectiveness without excessive complexity.
Data Source
AI summary
A multi-stage filtering process and system for fraud detection is disclosed. The multi-stage filtering process allows for the ability to dynamically increase/decrease server capacity, as well as application capacity to support fraud detection activities. The system monitors the queues of the transactions being made using various channels, and responds by adjusting the server and application resources needed for performing pre-filtering on the transactions in the queues. The invention allows for the fraud detection systems to maintain the capacity necessary to examine, at some level, if one or more of the transactions being processed by a financial institution are potentially fraudulent. The capacity can be changed during times of high and low volumes, thus allowing the allocation of resources based on transaction volume, which reduces the computing, energy, labor, etc. costs associated with fraud detection systems without losing the ability to detect almost all of the fraudulent transactions occurring.