Biometric Verification Using Predicted Signatures

• US 20150067822A1
• Filed: 09/26/2013
• Published: 03/05/2015
• Est. Priority Date: 09/05/2013
• Status: Active Grant

• Current Assignee: Barclays Execution Services Ltd. (Barclays PLC)
• Original Assignee: Barclays Bank PLC (Barclays PLC)
• Inventors: Randall, Lee
• Granted Patent: US 9,830,440 B2
• Time in Patent Office: Days
• US Class Current: 726/17
• CPC Class Codes: G06F 21/32 using biometric data, e.g. fingerprints, iris scans or voiceprints
• CPC Class Codes: G10L 17/04 Training, enrolment or model building
• CPC Class Codes: G10L 25/51 for comparison or discrimination

Abstract

A computer-implemented biometric identity verification method including the steps of storing a database of registered users, including data identifying profile attributes of each registered user and a respective plurality of stored biometric signatures, each stored biometric signature associated with a corresponding one or more of the profile attributes. A predicted biometric signature is derived for a requesting user when it is determined that a period of time has elapsed since the requesting user'"'"'s stored biometric signature was last updated, by adapting the stored biometric signature based on biometric variances derived from a biometric peer group of registered users with at least one profile attribute in common with the requesting user. The predicted biometric signature is used to verify the identity of the requesting user.

Claims

1. A computer-implemented biometric identity verification method, comprising the steps of:

storing a database of registered users on a computing device, the database including data identifying profile attributes of each registered user and a respective plurality of stored biometric signatures, each stored biometric signature associated with a corresponding one or more of the data identifying profile attributes;

deriving a new biometric signature for a requesting user in response to a request for identity verification;

determining that a predefined period of time has elapsed since the requesting user'"'"'s stored biometric signature was last updated;

identifying a group of registered users with at least one profile attribute in common with the requesting user;

calculating at least one variance value based on the stored biometric signatures of the identified group of registered users;

adapting the stored biometric signature of the requesting user based on the calculated at least one variance value; and

comparing the new biometric signature with the adapted stored biometric signature to verify the identity of the requesting user.

2. The method of claim 1, wherein the stored biometric signature includes one or more facial profiles.

3. The method of claim 1, wherein the stored biometric signature includes one or more voice prints.

4. The method of claim 3, wherein the one or more voice prints are derived from captured voice samples.

5. The method of claim 1, wherein the stored biometric signature comprises an index vector including a list of numerical code elements, each code element corresponding to an aspect of the data identifying profile.

6. The method of claim 5, wherein a list of variance values is calculated based on a respective plurality of index vectors of the stored biometric signatures that are retrieved for each registered user in the identified group of registered users.

7. The method of claim 6, wherein the variance values are calculated as a mean distance between corresponding code elements of the respective plurality of index vectors.

8. The method of claim 7, wherein the retrieved plurality of index vectors for each registered user in the identified group of registered users comprises a first index vector with at least one attribute matching the new biometric signature of the requesting user, and a second index vector with at least one attribute matching the present profile of the requesting user.

9. The method of claim 1, further comprising the step of receiving data representative of one or more present profile attributes from the requesting user.

10. The method of claim 8, wherein the stored biometric signature of the requesting user is adapted by combining respective variance values from the list of variance values with corresponding code elements in the index vector of the stored biometric signature.

11. The method of claim 1, further comprising the step of processing the database of registered users to determine a plurality of discrete groups of registered users having at least one profile attribute in common.

12. The method of claim 11, further comprising storing classification data identifying the determined plurality of discrete groups of registered users.

13. The method of claim 11, wherein identifying the group of registered users comprises identifying one or more of the determined plurality of discrete groups that shares a greatest number of profile attributes with the requesting user.

14. The method of claim 1, further comprising the step of storing the new biometric signature in the database after the identity of the requesting user is verified.

15. The method of claim 1, further comprising the step of comparing the new biometric signature with biometric signatures of a predefined group of known fraudsters.

16. The method of claim 1, wherein verifying the identity of the requesting user comprises calculating a distance measure between the new biometric signature with the adapted biometric signature, and determining that the calculated distance is within a predetermined threshold.

17. The method of claim 16, wherein the distance measure comprises a Euclidean distance, Hamming distance, a Bray-Curtis dissimilarity, a L₁ (Manhattan) or a Jaccard index.

18. The method of claim 16, wherein the predetermined threshold is tuned based on the calculated at least one variance value.

19. The method of claim 1, further comprising the step of identifying the last time the requesting user contacted the system.

20. The method of claim 1, further comprising the step of iteratively updating a registered user'"'"'s biometric signature based on voice-based interactions by the registered user with the system.

21. The method of claim 1, further comprising the step of verifying the identity of a registered user based on a six factor authentication consisting of identification data representative of something the system predicts, something the system finds out, something the system infers, something the user is, something the user knows, and something the user has.

22. The method of claim 1, further comprising the step of handling interactions with the computing device via a network gateway module and handling interactions with a telephone via a computer telephony integration module.

23. The method of claim 1, wherein the requesting user is a registered user.

24. A system for computer-implemented biometric identity verification, comprising:

a biometric profile database of registered users stored on a computing device, the biometric profile database configured to include data identifying profile attributes of each registered user and a respective plurality of stored biometric signatures, each stored biometric signature associated with a corresponding one or more of the data identifying profile attributes;

a customer relationship management database storing account data associated with a plurality of user accounts of the registered users;

a verification module performs verification of an account holder identity based on the stored biometric signatures associated with the account data of the registered users;

the verification module including;

a voice print comparator that receives a new biometric signature of a registered user of the system;

a group determiner associated with the customer relationship management database for identifying a group of registered users identified with at least one profile attribute in common with the requesting user;

a variance calculator configured to determine at least one variance value based on stored biometric signatures of the identified group of registered users and create an adapted biometric signature of the requesting user based on the at least one variance value; and

a comparator comparing the new biometric signature with the adapted biometric signature to verify the identity of the requesting user.

25. The system of claim 24, wherein the stored biometric signatures of the biometric profile database are selected from the group consisting of one or more facial profiles, and one or more voice prints.

26. The system of claim 25, wherein the one or more voice prints are derived from captured voice samples.

27. The system of claim 25, wherein the stored biometric signatures of the biometric profile database comprises an index vector including a list of numerical code elements, each code element corresponding to an aspect of the data identifying profile.

28. The system of claim 24, wherein the at least one variance value is calculated based on a respective plurality of index vectors of the stored biometric signatures that are retrieved for each registered user in the identified group of registered users.

29. The system of claim 28, wherein the at least one variance value is calculated as a mean distance between corresponding code elements of the respective plurality of index vectors.

30. The system of claim 29, wherein the respective plurality of index vectors for each registered user in the identified group of registered users comprises a first index vector with at least one attribute matching the new biometric signature of the requesting user, and a second index vector with at least one attribute matching the present profile of the requesting user.

31. The system of claim 30, wherein the stored biometric signature, from the biometric profile database, of the requesting user is adapted by combining respective variance values from the list of variance values with corresponding code elements in the index vector of the stored biometric signature.

32. The system of claim 24, further comprising receiving data representative of one or more present profile attributes from the requesting user.

33. The system of claim 24, wherein the group determiner in conjunction with the customer relationship management database determine a plurality of discrete groups of registered users having at least one profile attribute in common.

34. The system of claim 33, further comprising storing classification data identifying the determined plurality of discrete groups of registered users.

35. The system of claim 24, wherein the verification module compares the new biometric signature with biometric signatures of a predefined group of known fraudsters.

36. The system of claim 24, wherein the verification module calculates a distance measure between the new biometric signature and the adapted biometric signature, and determines that the calculated distance is within a predetermined threshold.

37. The system of claim 24, further comprising identifying the last time the requesting user contacted the system.

38. The system of claim 24, further including a voice print adapter updating a registered user'"'"'s biometric signature based on voice-based interactions by the registered user with the system.