System and method for biometric identification
Computer implemented systems and methods for matching an encrypted biometric input record with at least one stored encrypted biometric record, and without data decryption of the input and the at least one stored record.
1. A computer implemented method for matching an encrypted biometric input record with at least one stored encrypted biometric record, without data decryption of the input and the at least one stored record, the method comprising:
providing an initial biometric vector to a neural network, wherein the neural network translates the initial biometric vector to a Euclidian measurable feature vector;
storing the Euclidian measurable feature vector in a storage with other Euclidian measurable feature vectors;
classifying the Euclidian measurable feature vector; and/or classifying the current Euclidian measurable feature vector, wherein the classifying is performed at least in part using one or more distance functions;
receiving, from a mobile computing device over a data communication network, a current biometric vector representing the encrypted biometric input record;
providing the current biometric vector to the neural network, wherein the neural network translates the current biometric vector to a current Euclidian measurable feature vector; and
conducting a search of at least some of the stored Euclidian measurable feature vectors in a portion of the data storage using the current Euclidian measurable feature vector,
wherein the encrypted biometric input record is matched with at least one encrypted biometric record in encrypted space as a function of an absolute distance computed between the current Euclidian measurable feature vector and a calculation of each of the respective Euclidian measurable feature vectors in the portion of the storage,
wherein classifying the Euclidian measurable feature and/or the current Euclidian measurable feature vector returns floating point values, and a Frobenius algorithm is utilized to compute an absolute distance between each floating point and its average.
2. The method of claim 1 , wherein the search is conducted in Order log(n) time.
3. The method of claim 1 , further comprising:
using a Frobenius algorithm to classify the Euclidian measurable biometric vectors;
traversing a hierarchy of the classified Euclidian measurable biometric vectors in Order log(n) time; and
identifying that a respective Euclidian measurable biometric vector is the current Euclidian measurable feature vector.
4. The method of claim 1 , further comprising:
identifying, for each respective Euclidian measurable biometric vector, a plurality of floating point values; and
using a bitmap to eliminate from an absolute distances calculation any of the plurality of values that are not present in every vector.
5. The method of claim 1 , further comprising:
identifying, for each respective Euclidian measurable biometric vector, a plurality of floating point values; and
defining a sliding scale of importance based on the number of vectors a respective one of the floating point value appears.
6. The method of claim 1 , wherein the neural network is configured with a variety of convolutional layers, together with a rectifier (ReLU) and pooling nodes.
7. The method of claim 6 , further comprising:
computing, for each of a plurality of stored Euclidian measurable feature vectors, a relative position difference between an average face vector and the respective Euclidian measurable feature vector;
squaring the relative position difference;
summing the values; and
calculating the square root.
8. The method of claim 1 , wherein the neural network is configured to use pooling as a form of non-linear down-sampling,
and further wherein one or more pooling nodes progressively reduce the spatial size of a represented Euclidean-measurable feature vector to reduce the amount of parameters and computation in the neural network.
9. The method of claim 1 , wherein performance of the neural network is determined as a function of a cost function, in which a number of layers given as a spatial size of an output volume is computed as a function of an input volume size W, a kernel field size of layer neurons K, a stride with which the layers are applied S, and an amount of zero padding P used on a border.
10. The method of claim 1 , wherein the neural network translates the initial biometric vector and the current biometric vector as a function of matrix multiplications for each respective layer, and uses a Euclidean distance algorithm based on a Euclidean cost function.
11. A computer implemented system for matching an encrypted biometric input record with at least one stored encrypted biometric record, without data decryption of the input and the at least one stored record, the system comprising:
one or more processors and a computer-readable medium, wherein the one or more processors are configured to interact with the computer-readable medium in order to perform operations that include:
providing an initial biometric vector to a neural network, wherein the neural network translates the initial biometric vector to a Euclidian measurable feature vector;
storing the Euclidian measurable feature vector in a storage with other Euclidian measurable feature vectors;
receiving, from a mobile computing device over a data communication network, a current biometric vector representing the encrypted biometric input record;
classifying the Euclidian measurable feature vector; and/or classifying the current Euclidian measurable feature vector, wherein the classifying is performed at least in part using one or more distance functions;
providing the current biometric vector to the neural network, wherein the neural network translates the current biometric vector to a current Euclidian measurable feature vector; and
conducting a search of at least some of the stored Euclidian measurable feature vectors in a portion of the data storage using the current Euclidian measurable feature vector,
wherein the encrypted biometric input record is matched with at least one encrypted biometric record in encrypted space as a function of an absolute distance computed between the current Euclidian measurable feature vector and a calculation of each of the respective Euclidian measurable feature vectors in the portion of the storage,
wherein classifying the Euclidian measurable feature and/or the current Euclidian measurable feature vector returns floating point values, and a Frobenius algorithm is utilized to compute an absolute distance between each floating point and its average.
12. The system of claim 11 , wherein the search is conducted in Order log(n) time.
13. The system of claim 11 , wherein the one or more processors are configured to interact with the computer-readable medium in order to perform operations that include:
using a Frobenius algorithm to classify the Euclidian measurable biometric vectors;
traversing a hierarchy of the classified Euclidian measurable biometric vectors in Order log(n) time; and
identifying that a respective Euclidian measurable biometric vector is the current Euclidian measurable feature vector.
14. The system of claim 11 , wherein the one or more processors are configured to interact with the computer-readable medium in order to perform operations that include:
identifying, for each respective Euclidian measurable biometric vector, a plurality of floating point values; and
using a bitmap to eliminate from an absolute distances calculation any of the plurality of values that are not present in every vector.
15. The system of claim 11 , wherein the one or more processors are configured to interact with the computer-readable medium in order to perform operations that include:
identifying, for each respective Euclidian measurable biometric vector, a plurality of floating point values; and
defining a sliding scale of importance based on the number of vectors a respective one of the floating point value appears.
16. The system of claim 11 , wherein the neural network is configured with a variety of convolutional layers, together with a rectifier (ReLU) and pooling nodes.
17. The system of claim 16 , wherein the one or more processors are configured to interact with the computer-readable medium in order to perform operations that include:
computing, for each of a plurality of stored Euclidian measurable feature vectors, a relative position difference between an average face vector and the respective Euclidian measurable feature vector;
squaring the relative position difference;
summing the values; and
calculating the square root.
18. The system of claim 11 , wherein the neural network is configured to use pooling as a form of non-linear down-sampling,
and further wherein one or more pooling nodes progressively reduce the spatial size of a represented Euclidean-measurable feature vector to reduce the amount of parameters and computation in the neural network.
19. The system of claim 11 , wherein performance of the neural network is determined as a function of a cost function, in which a number of layers given as a spatial size of an output volume is computed as a function of an input volume size W, a kernel field size of layer neurons K, a stride with which the layers are applied S, and an amount of zero padding P used on a border.
20. The system of claim 11 , wherein the neural network translates the initial biometric vector and the current biometric vector as a function of matrix multiplications for each respective layer, and uses a Euclidean distance algorithm based on a Euclidean cost function.