Systems and methods for biometric processing with liveness
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 system for evaluating privacy-enabled biometrics for identification or authentication, the system comprising:
at least one processor operatively connected to a memory, the at least one processor when executing is configured to:
process an input including one or more plaintext instance of a first biometric data type, the input including the one or more plaintext instance of the first biometric data type and at least a candidate set of plaintext instances of another biometric data type capable of identification of an entity;
analyze a liveness threshold, wherein analyze the liveness threshold includes operations to determine that the candidate set of plaintext instances matches a random set of instances communicated to the entity to be identified or authenticated;
instantiate at least a first neural network trained to generate one way homomorphic encrypted feature vectors from the one or more plaintext instance of the first biometric type;
accept as an input to the first neural network at least the one or more plaintext instance of the first biometric type to output the one way homomorphic encrypted feature vectors of the first biometric type, and
generate a match to an identity based on the one way homomorphic encrypted feature vectors to one of a plurality of identification classes defined by the one way homomorphic encrypted feature vectors of the first biometric data type, or return an unknown result on a failure to match; and
confirm the match based, at least in part, on meeting the liveness threshold.
2. The system of claim 1 , wherein the at least one processor is further configured to instantiate the first neural network as at least a first pre-trained neural network trained on a first set of the plaintext instances of the first biometric type, the first pre-trained neural network once trained configured to generate the one way homomorphic encrypted feature vectors upon input of any plaintext instances of the first biometric type to the first pre-trained neural network.
3. The system of claim 1 , wherein the at least one processor is further configured to:
transform at least the candidate set of plaintext instances of the another biometric data type into a second format not configured for identification of the entity, and
determine that the candidate set of plaintext instances matches the random set of instances based on the second format.
4. The system of claim 1 , wherein the at least one processor is configured to select the candidate set of plaintext instances and request the entity seeking authentication or identification match the selected set of candidate plaintext instances.
5. The system of claim 1 , wherein the at least one processor is configured to assign a label to the entity to be identified or authenticated to enroll the entity.
6. The system of claim 5 , wherein the at least one processor is configured to return the label based on determining the match.
7. The system of claim 1 , wherein the at least one processor is configured to communicate or execute an application configured to:
initiate new input of plaintext instances of the first biometric data type;
generate new encrypted feature vectors from the new input of the plaintext instances of the first biometric type with a first pre-trained neural network; and
automatically delete the new input of the plaintext instances of the first biometric data type subsequent to generation of the new encrypted feature vectors.
8. A computer implemented method of evaluating privacy-enabled biometrics for identification or authentication and validity, the method comprising:
processing, by at least one processor, an input including one or more plaintext instance of a first biometric data type, the input including the one or more plaintext instance of the first biometric data type and at least a candidate set of plaintext instances of another biometric data type capable of identification of an entity;
analyzing, by the least one processor, a liveness threshold, wherein the act of analyzing the liveness threshold includes an act of determining, by the least one processor, that that the candidate set of plaintext instances matches a random set of instances communicated to the entity to be identified or authenticated;
instantiating, by the least one processor, at least a first neural network trained to generate one way homomorphic encrypted feature vectors from input plaintext instances of the first biometric type;
accepting as an input by the first neural network at least the one or more plaintext instance of the first biometric type to output the one way homomorphic encrypted feature vectors of the first biometric type;
generating, by the at least one processor, a match to an identity based on the one way homomorphic encrypted feature vectors to one of a plurality of identification classes defined by the one way homomorphic encrypted feature vectors of the first biometric data type, or return an unknown result on a failure to match; and
confirming, by the at least one processor, the match based, at least in part, on meeting the liveness threshold.
9. The method of claim 8 , wherein the method further comprises instantiating, by the at least one processor, the first neural network as at least a first pre-trained neural network trained on a first set of the plaintext instances of the first biometric type, and generating, by the first pre-trained neural network the one way homomorphic encrypted feature vectors upon input of any plaintext instances of the first biometric type to the first pre-trained neural network.
10. The method of claim 9 , wherein the first pre-trained neural network is configured to generate the one way homomorphic encrypted feature vectors independent of an identity of the entity to be identified or authenticated.
11. The method of claim 8 , wherein the method further comprises:
transforming, by the at least one processor, at least the candidate set of plaintext instances of the another biometric data type for identification of the entity into a second format not configured for identification of the entity, and
determining, by the at least one processor, that the candidate set of plaintext instances matches the random set of instances based on the second format.
12. The method of claim 8 , wherein the method further comprises assigning a label to the entity to be identified or authenticated to enroll the entity.
13. The method of claim 12 , wherein the method further comprises returning the label based on determining the match.
14. The method of claim 8 , wherein the method further comprises:
initiating, by the at least one processor, new input of plaintext instances of the first biometric data type;
generating, by the at least one processor, new encrypted feature vectors from the new input of the plaintext instances of the first biometric type with a first pre-trained neural network; and
automatically deleting, by the at least one processor, the new input of the plaintext instances of the first biometric data type subsequent to generation of the new encrypted feature vectors.
15. A non-transitory computer readable medium containing instructions that when executed cause at least one processor to perform a method of evaluating privacy-enabled biometrics for identification or authentication and validity, the method comprising:
processing an input including one or more plaintext instance of a first biometric data type, the input including the one or more plaintext instance of the first biometric data type and at least a candidate set of plaintext instances of another biometric data type capable of identification of an entity;
analyzing a liveness threshold, a liveness threshold, wherein the act of analyzing the liveness threshold includes an act of determining that that the candidate set of plaintext instances matches a random set of instances communicated to the entity to be identified or authenticated;
instantiating at least a first neural network trained to generate one way homomorphic encrypted feature vectors from input plaintext instances of the first biometric type;
accepting as an input by the first neural network at least some of the one or more plaintext instances of the first biometric type to output the one way homomorphic encrypted feature vectors of the first biometric type;
generating a match to an identity based on the one way homomorphic encrypted feature vectors to one of a plurality of identification classes defined by the one way homomorphic encrypted feature vectors of the first biometric data type, or return an unknown result on a failure to match; and
confirming the match based, at least in part, on meeting the liveness threshold.
16. The medium of claim 15 , wherein the method further comprises instantiating, by the at least one processor, the first neural network as at least a first pre-trained neural network trained on a first set of the plaintext instances of the first biometric type, and generating, by the first pre-trained neural network the one way homomorphic encrypted feature vectors upon input of any plaintext instances of the first biometric type to the first pre-trained neural network.
17. The medium of claim 16 , wherein the first pre-trained neural network is configured to generate the one way homomorphic encrypted feature vectors independent of an identity of the entity to be identified or authenticated.
18. The medium of claim 15 , wherein the method further comprises:
transforming the at least the candidate set of plaintext instances of the another biometric data type into a second format not configured for identification of the entity; and
determining that the candidate set of plaintext instances matches the random set of instances based on the second format.
19. The medium of claim 15 , wherein the method further comprises assigning a label to the entity to be identified or authenticated to enroll the entity.
20. The medium of claim 19 , wherein the method further comprises returning the label based on determining the match.