Biometric Authentication Using Local Neural Network Weights
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional biometric authentication systems require a central database for comparison, leading to extensive communication infrastructure needs, security risks, and user data transmission over networks, as well as the need for users to share biometric information with multiple service providers.
Innovation Solution
A method using an authentication device with untrained artificial neural networks that receives a data set from a user device, generates trained networks, and determines the correlation between biometric attributes of a reference user and a test user, allowing for local authentication without a central database, using encrypted data sets and various biometric attributes like voice prints, fingerprints, and images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a central database is implemented for biometric authentication, then user verification can be performed across multiple access points, but extensive communication infrastructure is required and security risks increase due to data transmission over networks
Solution Approach 1:
The patent extracts the biometric template from the central database and places it locally in the authentication device. Only the extracted template is stored and processed locally, eliminating the need to transmit full biometric data over networks while maintaining authentication capability across multiple access points.
Solution Approach 2:
The system segments the authentication process into two parts: biometric template extraction and local verification. The extracted template is stored separately in the authentication device, allowing verification to occur locally without requiring continuous communication with a central database, thus reducing network dependency and security risks.
2Extent of automation
If biometric information is transmitted over communication networks for every authentication event, then centralized verification is possible, but encryption overhead and security risks increase
Solution Approach 1:
The system performs preliminary extraction of the biometric template during an initial enrollment phase. This extracted template is then stored locally in the authentication device, eliminating the need for repeated transmission and encryption of biometric data during subsequent authentication events, thus reducing encryption overhead.
3Device complexity
If a central database stores biometric data, then user enrollment is centralized, but users must provide biometric profiles to every service provider separately
Solution Approach 1:
The authentication device is designed with universal functionality to store and verify extracted biometric templates. This universal design allows the same device to work across different service providers and access points without requiring separate enrollment at each location, enhancing service provider compatibility.
Data Source
AI summary
Methods and apparatuses for providing biometric authentication of a test user using a registration process where a reference data sample representative of one or more biometric attributes of a reference user is used to train a plurality of neural networks to achieve a target output. The weights that achieve this in the plurality of neural networks may be stored on a user device as a first data set. A second data set representative of one or more biometric attributes may be obtained from the test user using the authentication device or received from the user device. The first data set may be received by the authentication device and used as weights in an artificial neural network and the second data set may be used as inputs. The output of the neural network may determine a degree of correlation between the reference user and the test user to be authenticated.