Biometric Data Classification for Authentication Database Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The performance of deep learning-based face recognition and authentication systems is dependent on the composition of the training and performance test databases, and determining which characteristic information to store in these databases to improve performance is challenging.
Innovation Solution
A biometric authentication data classification device and method that uses an artificial neural network model to extract and calculate the overall similarity between candidate biometric data and performance test data, allowing for the addition of candidate data to training or performance test databases based on similarity thresholds or distribution ratios, thereby improving database composition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If deep learning models are applied to face image authentication, then recognition performance can be improved, but the performance becomes dependent on the training database composition making it difficult to determine which characteristic information to store
Solution Approach 1:
The patent segments biometric data into multiple characteristic dimensions (age, gender, race, facial shape, etc.) and evaluates each dimension independently using similarity calculations. This allows systematic determination of which specific characteristic segments should be stored in the database to improve authentication performance without overwhelming complexity.
Solution Approach 2:
The patent changes the approach from storing raw biometric data to storing extracted characteristic parameters (age, gender, race, facial shape features). By transforming the data representation to parameter-based characteristics, the system can systematically select and store only the most relevant parameters that improve authentication performance.
2Measurement precision
If the training database is improved to enhance recognition performance, then authentication accuracy increases, but determining which specific characteristic information to include becomes challenging
Solution Approach 1:
The patent implements a feedback mechanism where candidate biometric data is evaluated by calculating overall similarity across multiple characteristic dimensions. The system provides feedback on which characteristics contribute most to recognition accuracy, allowing iterative refinement of the training database composition based on measured performance improvements.
Solution Approach 2:
The patent performs preliminary extraction and evaluation of characteristic information from candidate biometric data before final database construction. By pre-calculating similarity metrics and characteristic importance for candidate data, the system determines in advance which specific characteristics should be included in the training database to achieve target recognition accuracy.
3Reliability
If performance test databases are used to evaluate authentication systems, then performance can be measured, but databases with low performance make it difficult to identify useful characteristic information
Solution Approach 1:
The patent converts the problem of using low-performance test databases into an opportunity. Instead of discarding poor-performing test data, the system calculates overall similarity for candidate data against these low-performance databases, identifying which characteristic dimensions, when improved, would yield the greatest performance gains. This transforms weak test results into valuable guidance for database optimization.
Data Source
AI summary
A method for adding biometric authentication training data into databases performed by a biometric authentication data classification device includes: extracting first biometric characteristic information from at least one candidate biometric training data for biometric authentication using an artificial neural network model; calculating an overall similarity between the first biometric characteristic information and second biometric characteristic information extracted from a performance test database of which a biometric authentication performance is lower than a threshold level, the performance test database being selected among performance test databases for the biometric authentication; and adding the at least one candidate biometric training data into one of the biometric authentication training database and the performance test database based on the calculated overall similarity.