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) 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. This improves security over conventional approaches. Searching biometrics in the clear on any system, represents a significant security vulnerability. In various examples described herein, only the one-way encrypted biometric data is available on a given device. Various embodiments restrict execution to occur on encrypted biometrics for any matching or searching.
1. A privacy-enabled biometric system comprising:
at least one processor operatively connected to a memory;
the at least one processor when executing configured to:
instantiate a classification component comprising a classification network that forms a first member of an operative pairing of networks based on biometric information type, wherein the classification network is configured to:
classify distance measurable encrypted feature vectors and label inputs for identification during training,
accept as an input encrypted feature vectors that are distance measurable and are produced at least from a one way encoding of plaintext identification information by a first pre-trained neural network forming a second member of the operative paring of networks, and
predict an outcome based on a trained model and a set of inputs for the prediction to match to a result label or unknown;
return an unknown result or a matching label from a plurality of trained identification classes as an output of prediction; and
wherein the at least one processor is configured to:
delete any plaintext biometric information responsive to generation of the distance measurable encrypted feature vectors by the first pre-trained neural network.
2. The system of claim 1 , wherein the at least one processor is configured to:
instantiate an enrollment interface configured to:
accept plaintext biometric information;
provide distance measurable encrypted feature vectors for classification generated at least in part from the first pre-trained neural network from the plaintext biometric information;
trigger deletion of the plaintext biometric information responsive to generation of the distance measurable encrypted feature vectors by the first pre-trained neural network; and
provide a respective label to the classification component for training with associated one or more distance measurable encrypted feature vectors; and
trigger the classification component to train the deep neural network on the distance measurable encrypted feature vectors to link the label to the associated encrypted feature vectors.
3. The system of claim 1 , wherein the enrollment interface includes a local component installed on a mobile device.
4. The system of claim 3 , wherein the at least one processor is configured to execute a local prediction operation on the mobile device with the deep neural network to match a newly generated distance measurable encrypted feature vector.
5. The system of claim 4 , wherein the at least one processor is configured to execute a remote prediction operation if the local operation returns unknown on the newly generated distance measurable encrypted feature vector.
6. The system of claim 5 , wherein the at least one processor is configured to update at least one of a remote copy of the deep neural network or a local copy of the deep neural network with the newly generated distance measurable encrypted feature vector.
7. The system of claim 5 , wherein the enrollment interface is accessible via uri, and is configured to accept unencrypted biometric information and personally identifiable information (“PII”).
8. The system of claim 7 , wherein the enrollment interface is configured to link the PII to a one way homomorphic encryption of an unencrypted biometric input.
9. The system of claim 1 , wherein the classification component is further configured to return a set of probabilities for matching a feature vector in an array of values.
10. The system of claim 1 , further comprising an interface configured to accept a biometric input and return and indication of known or unknown to a requesting entity.
11. The system of claim 10 , wherein requesting entity includes any one or more of: an application, a mobile application, a local process, a remote process, a method, and a business object.
12. The system of claim 1 , wherein the classification component further comprising multiple classification networks for processing different types of biometric information.
13. The system of claim 1 , wherein the classification component is further configured to match an identity of a subject responsive to at least two probable biometric indicators that may be used simultaneously or as part of a voting algorithm.
14. The system of claim 1 , wherein the system is further configured to enroll a subject for biometric identification or authentication; and the classification component is further configured to accept input of distance measurable encrypted feature vectors for subject identification during prediction.
15. A computer implemented method for using privacy-enabled biometrics, the method comprising:
instantiating, by at least one processor, a classification network having a classification network that forms a first member of an operative pairing of networks based on biometric information type;
classifying, by the classification network, distance measurable encrypted feature vectors and label inputs for identification during training;
accepting as an input, by the classification network, encrypted feature vectors that are distance measurable and are produced at least from a one way encoding of plaintext biometric information by a first pre-trained neural network forming a second member of the operative paring of networks;
predicting, by the deep neural network, an outcome based on a trained model and a set of inputs for the prediction to match to a result of a label or unknown;
returning an unknown result or a matching label from a plurality of trained identification classes as an output of prediction; and
deleting, by the at least one processor, any plaintext biometric information responsive to generation of the distance measurable encrypted feature vectors by the first pre-trained neural network.
16. The method of claim 15 , further comprising:
instantiating, by the at least one processor, an enrollment interface;
accepting, by the enrollment interface, plaintext biometric information;
providing distance measurable encrypted feature vectors for classification generated at least in part from the first pre-trained neural network from the plaintext biometric information;
triggering the act of deleting any of the plaintext biometric information responsive to generation of the distance measurable encrypted feature vectors by the first pre-trained neural network;
providing a respective label to the classification component for training with associated one or more distance measurable encrypted feature vectors; and
triggering the classification component to train the classification network on the distance measurable encrypted feature vectors to link the label to the associated encrypted feature vectors.
17. The method of claim 16 , wherein act of accepting the plaintext biometric information is executed by a mobile device.
18. The method of claim 17 , wherein the method further comprises executing a local prediction operation on the mobile device with the deep neural network to match a newly generated distance measurable encrypted feature vector.
19. The method of claim 18 , wherein the method further comprises executing a remote prediction operation if the local operation returns unknown on the newly generated distance measurable encrypted feature vector.
20. The method of claim 19 , wherein the method further comprises updating at least one of a remote copy of the classification network or a local copy of the classification network with the newly generated distance measurable encrypted feature vector.