Facial Recognition Fraud Detection via Bypass Information Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current face recognition systems are vulnerable to fraudulent behaviors, such as using non-live objects like photos or videos, which existing technologies fail to detect accurately and reliably.
Innovation Solution
A method and apparatus that collect device and user behavior information, known as bypass information, to input into decision models like decision trees or SVMs to predict the probability of fraudulent behavior, improving detection accuracy by using stable bypass information that is less affected by environmental factors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional face recognition methods are used, then the face recognition process is simple and fast, but the system is vulnerable to fraudulent behaviors such as photo or video attacks
Solution Approach 1:
The patent segments the fraud detection process into multiple independent components: device information collection, user behavior information collection, and decision model analysis. Each component handles a specific aspect of bypass information, making the complex detection system manageable and modular while improving reliability against fraudulent behaviors
Solution Approach 2:
The patent introduces decision models as intermediary components that process bypass information and provide fraud probability predictions. These decision models act as mediators between the collected bypass information and the final fraud determination, enhancing detection accuracy while maintaining system organization
2Measurement precision
If bypass information collection is added to detect fraudulent behaviors, then detection accuracy improves, but processing time increases
Solution Approach 1:
The patent collects bypass information (device information and user behavior information) in advance during the face recognition process, before the final fraud determination is made. This preliminary collection allows the decision models to quickly analyze pre-gathered data, improving detection accuracy without adding significant processing time to the critical recognition path
Solution Approach 2:
The patent implements a multi-level decision model approach where simple rules can quickly resolve obvious cases, while more complex analysis is applied only when needed. This partial application of comprehensive bypass information analysis maintains high detection accuracy while reducing average processing time for non-fraudulent cases
Data Source
AI summary
A computer-implemented method for detecting fraudulent behavior in a facial recognition process includes: receiving, by a computing device, a facial recognition request from a user; collecting bypass information of the user, in which the bypass information includes user device information and user behavior information; inputting the bypass information into at least one decision model to obtain a bypass decision result; and determining, based on the bypass decision result, whether fraudulent behavior is present in the facial recognition process.