Adaptive Biometric Anti-Spoofing via Online Model Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Biometric authentication systems are vulnerable to spoofing attacks, where fake biometric data is used to gain unauthorized access, due to the inability of existing anti-spoofing protection models to accurately distinguish between real and fake biometric data.
Innovation Solution
The implementation of an adaptive personalization method using online data, where a biometric data input is processed through a first machine learning model to extract features, and a second model determines its authenticity. The model is then refined using a finetuning data set, with features and labels added based on predictive scores above or below certain thresholds, to improve the accuracy of anti-spoofing protection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional biometric authentication systems are used, then user convenience is improved, but vulnerability to spoofing attacks increases
Solution Approach 1:
The patent introduces an anti-spoofing protection model as an intermediary layer between the biometric authentication system and the spoofing attacks. This model analyzes biometric data to detect signs of spoofing while allowing legitimate authentication to proceed smoothly, thus maintaining user convenience while improving security against spoofing attacks
Solution Approach 2:
The patent changes the parameters of biometric data analysis by extracting additional features and using multiple machine learning models with different parameters to detect spoofing. By analyzing multiple parameters of biometric data simultaneously, the system can distinguish between real and fake biometric inputs without significantly impacting user convenience
2Reliability
If anti-spoofing protection models are added to biometric authentication systems, then security against spoofing is improved, but system complexity increases
Solution Approach 1:
The patent segments the anti-spoofing protection into multiple independent machine learning models, each responsible for specific aspects of spoofing detection. This segmentation allows the system to achieve comprehensive security while keeping each individual model relatively simple and manageable
Solution Approach 2:
The patent designs machine learning models that serve multiple functions: they not only detect spoofing but also improve their accuracy over time by learning from both authentic and spoofed biometric data. This multi-functionality reduces the need for separate systems for detection and learning, thereby reducing overall system complexity
3Measurement precision
If machine learning models are used for anti-spoofing detection, then detection accuracy is improved, but computational requirements increase
Solution Approach 1:
The patent applies partial action by using multiple machine learning models with increasing complexity only when needed for spoofing detection. For legitimate authentication cases, simpler verification processes are used, reducing computational requirements while maintaining high detection accuracy when spoofing is suspected
Data Source
AI summary
Certain aspects of the present disclosure provide techniques and apparatus for biometric authentication using an anti-spoofing protection model refined using online data. The method generally includes receiving a biometric data input for a user. Features for the received biometric data input are extracted through a first machine learning model. It is determined, using the extracted features for the received biometric data input and a second machine learning model, whether the received biometric data input for the user is authentic or inauthentic. It is determined whether to add the extracted features for the received biometric data input, labeled with an indication of whether the received biometric data input is authentic or inauthentic, to a finetuning data set. The second machine learning model is adjusted based on the finetuning data set.