Automated Feature Generation for Identity Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing identity verification and fraud detection systems face challenges in timeliness and accuracy due to reliance on static databases, which can lead to stale data and increased costs from third-party data sources, resulting in inadequate decision-making for online transactions.
Innovation Solution
A system employing novel machine learning and feature engineering techniques to create dynamic data sets for transaction decision-making, using feedback from previous transactions to generate feature sets that assess fraud and identity verification risks in real-time, optimizing decision-making through cloud-based applications and APIs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If static databases are used for identity verification and fraud detection, then system simplicity is maintained, but data timeliness and accuracy deteriorate
Solution Approach 1:
The patent transforms static databases into dynamic systems by implementing automated feature generation that continuously creates new data fields from transaction information. The system dynamically generates features like device fingerprints, location data, and behavioral patterns in real-time, allowing the database to adapt and update automatically without manual intervention, thus resolving the contradiction between maintaining system simplicity and achieving data timeliness.
Solution Approach 2:
The system implements feedback mechanisms where transaction outcomes (fraudulent or legitimate) are fed back into the database to continuously refine and update feature sets. This closed-loop approach ensures that the database learns from new information and updates its verification criteria in real-time, improving accuracy while maintaining operational simplicity through automated learning rather than manual updates.
2Measurement precision
If third-party data sources are extensively used for fraud detection, then assessment accuracy improves, but system costs increase
Solution Approach 1:
The patent enables the system to generate its own verification features from internal transaction data without relying on expensive third-party sources. By implementing automated feature generation that extracts meaningful patterns from existing transaction information, device data, and user behavior, the system becomes self-sufficient in creating accurate fraud indicators, thereby reducing external data purchase costs while maintaining assessment precision.
Solution Approach 2:
The system changes the parameters being measured by creating new derived features from existing data. Instead of purchasing external data, the system transforms internal transaction parameters into meaningful fraud indicators through automated feature generation, effectively changing what parameters are monitored and how they are derived, thus reducing costs while maintaining or improving detection accuracy.
3Reliability
If manual database updates are performed for fraud detection, then data accuracy can be controlled, but productivity and timeliness deteriorate
Solution Approach 1:
The system performs self-updates through automated feature generation that continuously creates and refines data fields without human intervention. The automated processes extract features from incoming transactions, update the database in real-time, and maintain data accuracy through algorithmic consistency rather than manual review, thereby achieving both high productivity and reliable data accuracy simultaneously.
Solution Approach 2:
The patent replaces manual mechanical update processes with automated computational systems. Instead of human operators manually updating databases, the system uses automated feature generation algorithms that computationally create and update data fields in real-time, substituting mechanical human labor with automated computational processes that maintain accuracy while dramatically improving update speed and productivity.
4Reliability
If comprehensive transaction data is collected for fraud detection, then detection capability improves, but device complexity increases
Solution Approach 1:
The patent segments the comprehensive transaction data into distinct feature categories that are generated and processed independently. By dividing the data collection and processing into modular feature generation components (device features, transaction features, behavioral features), the system manages complexity through segmentation while still utilizing comprehensive data for improved fraud detection capability.
Solution Approach 2:
The automated feature generation system serves multiple functions simultaneously: it collects comprehensive transaction data, processes and transforms it into meaningful features, stores it in the database, and uses it for fraud detection. This multi-functional approach reduces overall system complexity by having a single universal mechanism handle multiple tasks rather than requiring separate systems for each function.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The system and methodology of the present invention employs available data obtained in connection with previous transactions to create one or more databases comprising feature sets which are used in transaction decision making solutions. The available data from previous transactions which is used in creating feature sets may include all available production data or the data may be stratified across specific industries and/or across specific decision support customers to optimize the expected decision making results. A feature engine is provided which uses a combination of data, time and combinational aggregate functions to feature engineer one or more feature sets used for one or more purposes, such purposes to include identity verification, fraud assessment, document verification as well as other assessments related to selectively permit or not permit transactions to proceed.