Dynamic Authorization Thresholds for Facial Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern noncash transactions are vulnerable to identity theft, as dishonest individuals can misuse credit or debit card information, leading to unauthorized charges, and existing verification methods lack robustness in distinguishing genuine cardholders, especially for higher-value transactions.
Innovation Solution
A system that uses facial recognition to generate a confidence value by comparing a user's photo to reference photos stored in a database, with an authorization threshold set based on the transaction value, ensuring higher confidence levels for larger transactions, thereby reducing unauthorized transactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional verification methods are used for noncash transactions, then transaction speed is maintained, but security against identity theft deteriorates
Solution Approach 1:
The patent replaces traditional mechanical verification methods (manual signature verification, visual ID checking) with an automated facial recognition system using biometric technology. The analysis engine compares the captured user photo with reference photos from the database to generate a confidence value, substituting human verification with automated biometric authentication to enhance security while maintaining transaction efficiency
Solution Approach 2:
The patent introduces a confidence value as an intermediary metric between the facial recognition comparison and the authorization decision. This intermediary allows for flexible threshold setting based on transaction values, enabling the system to balance security and convenience dynamically without requiring complex rule-based decision systems
2Reliability
If higher authorization thresholds are set for all transactions, then security is improved, but transaction approval rate deteriorates
Solution Approach 1:
The patent makes the authorization threshold dynamic by setting different thresholds based on transaction values. Lower thresholds are applied to smaller transactions where the risk is lower, while higher thresholds are applied to larger transactions. This dynamic approach maintains high security for high-value transactions while preserving convenient approval rates for routine low-value transactions
Solution Approach 2:
The patent changes the authorization threshold parameter based on the transaction value parameter. The system adjusts the confidence value requirement dynamically - for example, a $5 transaction may require 70% confidence while a $5000 transaction may require 95% confidence. This parameter adjustment resolves the contradiction by adapting security levels to the actual risk profile of each transaction
3Reliability
If facial recognition with confidence values is implemented, then security is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary actions by pre-storing reference photos of cardholders in the database during account setup. When a transaction occurs, the system only needs to compare the captured photo against these pre-prepared references rather than performing full identity verification from scratch. This preliminary preparation significantly reduces processing time during actual transactions while maintaining high verification accuracy
Data Source
AI summary
An embodiment of the invention provides a method for selectively using degree confidence for image validation to authorize transactions, wherein a request to authorize a transaction is received, the request including a user's name, a photo of the user, and a transaction value. A database is queried with the user's name to identify one or more reference photos of the user; and, the received photo of the user is compared to the reference photo of the user with an analysis engine to generate a confidence value. An authorization threshold is determined with a processor based on the transaction value. The transaction is authorized when the confidence value is equal to and/or greater than the authorization threshold.