Systems and methods for privacy-enabled biometric processing
In one embodiment, a set of feature vectors can be derived from any biometric data, and then using a deep neural network (“DNN”) on those one-way homomorphic encryptions (i.e., each biometrics' feature vector) an authentication system can determine matches or execute searches on encrypted data. Each biometrics' feature vector can then be stored and/or used in conjunction with respective classifications, for use in subsequent comparisons without fear of compromising the original biometric data. In various embodiments, the original biometric data is discarded responsive to generating the encrypted values. In another embodiment, the homomorphic encryption enables computations and comparisons on cypher text without decryption of the encrypted feature vectors. Security of such privacy enable biometrics can be increased by implementing an assurance factor (e.g., liveness) to establish a submitted biometric has not been spoofed or faked.
1. A privacy-enabled biometric system comprising:
at least one processor operatively connected to a memory, the at least one processor configured to:
establish an authentication mode of operation;
execute a first machine learning (“ML”) process based on the authentication mode, wherein the first ML process when executed by the at least one processor is configured to accept distance measurable encrypted feature vectors as input and classify the distance measurable encrypted feature vector input as part of identification or authentication of an entity using a first classification network trained on the distance measurable encrypted feature vectors, produced, at least in part, by a pre-trained generation neural network, for a plurality of identification classes, to determine a match to information of an identification data type;
execute a second ML process based on the authentication mode, wherein the second ML process when executed by the at least one processor is configured to:
compare distances between at least one stored distance measurable encrypted feature vector and a newly generated distance measurable encrypted feature vector of the identification data type, produced, at least in part, by the pre-trained generation neural network, during identification or authentication of an entity to determine a match; and
return a label associated with the entity identified by one or both of the first ML process or the second ML process, or return an unknown result on failure to match.
2. The system of claim 1 , wherein the second ML process when executed by the at least one processor is configured to process plain text identification inputs of the identification data type using the pre-trained neural network that is trained to generate distance measurable encrypted feature vectors from the plain text identification inputs of the identification data type.
3. The system of claim 1 , wherein at least one of the first ML process or the second ML process is configured, during enrollment, to:
determine one or more distances between encrypted feature vectors produced by the pre-trained generation neural network;
exclude encrypted feature vectors having one or more distances exceeding a threshold distance for subsequent training processes; and
include encrypted feature vectors having distances within the threshold distance for subsequent training processes.
4. The system of claim 1 , wherein the at least one processor is configured to determine the authentication mode and select one or both of the first ML process or the second ML process for execution.
5. The system of claim 3 , wherein the at least one processor is configured to trigger at least training operations of one or both the first and second ML processes responsive to determining that a current authentication mode includes an enrollment mode.
6. The system of claim 3 , wherein the at least one processor is configured to execute at least part of the second ML process to authenticate a new user until at least a period of time required for training the first classification network expires.
7. The system of claim 6 , wherein the at least one processor is configured to execute at least part of the first ML process to authenticate a new user responsive to completing training of the first classification network.
8. The system of claim 1 , wherein the first classification network comprises a deep neural network (“DNN”), wherein the DNN is configured to:
generate an array of values in response to input of at least one unclassified encrypted feature vector during authentication; and
determine a label or unknown result based on analyzing the generate array of values.
9. The system of claim 1 , wherein the first classification network and the pre-trained generation neural network define an operative pairing based on a first biometric data type, the operative pairing including at least one instance of a first DNN and at least one instance of a first pre-trained generation neural network that are collectively configured to process and predict matches on input of the first biometric data type.
10. The system of claim 9 , wherein the at least one processor is configured to instantiate at least a second operative pairing of neural networks including a second classification network and a second pre-trained generation neural network, the second operative pairing configured to process and predict matches on a second biometric data type.
11. The system of claim 1 , wherein the at least one processor is configured to communicate or execute an application configured to:
initiate an input of plaintext instances of a first biometric data type; and
automatically delete the plaintext instances of the first biometric data type subsequent to generation of the distance measurable encrypted feature vectors of the first biometric data type.
12. The system of claim 1 , wherein the first pre-trained generation neural network is configured to generate the distance measurable encrypted feature vectors during prediction, independent of an identity of the entity to be identified or authenticated.
13. A computer implemented method for privacy enabled authentication, the method comprising:
establishing an authentication mode of operation;
triggering, by at least one processor, one or both of a first machine learning (“ML”) process or a second ML process responsive to determining an authentication mode;
executing, by the at least one processor, the first ML process, wherein executing the first ML process includes:
accepting at least one distance measurable encrypted feature vector, produced, at least in part, by a pre-trained generation neural network, as input to a first classification-network, and classifying the distance measurable encrypted feature vector input as part of identification or authentication of an entity using the first classification network trained on the distance measurable encrypted feature vectors for a plurality of identification classes, to determine a match;
executing, by the at least one processor, the second ML process, wherein executing the second ML process includes:
comparing, by the at least one processor, distances between at least one stored distance measurable encrypted feature vector and a newly generated distance measurable encrypted feature vector during identification or authentication of the entity to determine a match, the newly generated distance measurable encrypted feature vector produced, at least in part, by the pre-trained neural network; and
returning a label associated with the entity identified by one or both of the first ML process or the second ML process, or returning an unknown result on failure to match.
14. The method of claim 13 , further comprising:
processing plain text identification inputs of an identification data type using a pre-trained neural network trained to generate distance measurable encrypted feature vectors from the plain text identification inputs of the identification data type.
15. The method of claim 13 , further comprising:
determining one or more distances between encrypted feature vectors produced by the pre-trained generation neural network;
excluding encrypted feature vectors having one or more distances exceeding a threshold distance for subsequent training processes; and
including encrypted feature vectors having distances within the threshold distance for subsequent training processes.
16. The method of claim 13 , further comprising determining the authentication mode and select one or both of the first ML process or the second ML process for execution.
17. The method of claim 13 , further comprising determining the authentication mode includes operations to identify an enrollment mode for establishing a new entity for subsequent authentication.
18. The method of claim 13 , wherein the first classification network and the pre-trained generation neural network define an operative pairing based on a first biometric data type, the operative pairing including at least one instance of a first DNN and at least one instance of a first pre-trained generation neural network that are collectively configured to process and predict matches on input of the first biometric data type.
19. The method of claim 18 , wherein the method further comprises instantiating at least a second operative pairing of neural networks including a second classification network and a second pre-trained generation neural network, the second operative pairing configured to process and predict matches on a second identification data type.
20. The method of claim 13 , wherein the method further comprises:
communicating or executing an application;
initiating, by the application, an input of plaintext instances of a first biometric data type; and
automatically deleting, by the application, the plaintext instances of the first biometric data type subsequent to generation of the distance measurable encrypted feature vectors of the first biometric data type.
21. The method of claim 13 , wherein the method further comprises generating by the first pre-trained generation neural network the distance measurable encrypted feature vectors during prediction, independent of an identity of the entity to be identified or authenticated.