User identity authentication techniques for on-line content or access

• US 10,095,850 B2
• Filed: 05/19/2015
• Issued: 10/09/2018
• Est. Priority Date: 05/19/2014
• Status: Active Grant

• Current Assignee: Kadenze, Inc.
• Original Assignee: Kadenze, Inc.
• Inventors: Cook, Perry R., Kapur, Ajay, Vallis, Owen S., Hochenbaum, Jordan
• Primary Examiner(s): Gracia, Gary S
• Application Number:
  US14/716,591
• Publication Number:
  US 20150379253A1
• Time in Patent Office: 1,239 Days
• Field of Search: 381 56, 726 16, 726 19, 434157, 434362, 713186
• CPC Class Codes: G06F 21/316 by observing the pattern of computer usage, e.g. typical user behaviour
• CPC Class Codes: G06F 21/32 using biometric data, e.g. fingerprints, iris scans or voiceprints
• CPC Class Codes: G06Q 40/12 Accounting
• CPC Class Codes: G06Q 50/205 Education administration or guidance
• CPC Class Codes: G09B 5/00 Electrically-operated educational appliances working with questions and answers G09B7/00; simulators G09B9/00; advertising or displaying in general G09F
• CPC Class Codes: G09B 5/06 with both visual and audible presentation of the material to be studied
• CPC Class Codes: G09B 7/02 of the type wherein the student is expected to construct an answer to the question which is presented or wherein the machine gives an answer to the question presented by a student
• CPC Class Codes: H04L 63/0861 using biometrical features, e.g. fingerprint, retina-scan cryptographic mechanisms or cryptographic arrangements for entity authentication using biological data H04L9/3231

Abstract

On-line course offerings can be made available to users using computational techniques that reliably authenticate the identity of individual student users during the course of the very submissions and/or participation that will establish student user proficiency with course content. Authentication methods and systems include applications of behavioral biometrics.

Claims

1. A multi-modal authentication method, comprising:

initially capturing an enrollment dataset of at least first- and second-type biometrics characteristic of a particular user;

subsequent to the initial capturing, simultaneously capturing by way of one or more computing device interfaces, (i) a first dataset corresponding to the first-type biometric and (ii) a second dataset corresponding to the second-type biometric, wherein the first-type biometrics include audio features extracted from vocals of the user,

wherein the second-type biometrics include visual features extracted from an image or video of the user, and

wherein the first and second datasets are coherent with each other, time-aligned and correspond to a same interactive response by the user to tracking of an on-screen moving target using an on-screen aiming mechanism for detecting and tracking facial landmarks employed to improve quality or uniformity of the image or video of the user;

computationally determining correspondence of the first- and second-type biometrics with the enrollment dataset;

authenticating an identity of the user based on the determined correspondence (i) between the first- and second-type biometrics with the enrollment dataset and (ii) between the time-aligned audio and visual features.

2. The multi-modal authentication method of claim 1, wherein the correspondence determining further includes:

computationally determining correspondence between time-aligned ones of the audio and visual features respectively extracted from the vocals and image or video of the user in connection with the interactive response.

3. The multi-modal authentication method of claim 1, wherein for the first-type biometrics, the correspondence determination includes comparison of Mel frequency cepstrum coefficients (MFCCs) or spectral subband centrolds (SSCs) using a trained classifier.

4. The multi-modal authentication method of claim 1, wherein for the second-type biometrics, the correspondence determination includes use of a local binary patterns histogram (LBPH) technique performed on an aligned, cropped and illumination adjusted image.

5. The multi-modal authentication method of claim 1, further comprising:

performing the capturing of the enrollment dataset using the one or more computing device interfaces and in connection with the user'"'"'s interactive response thereon.

6. The multi-modal authentication method of claim 1, wherein the first and second datasets are captured covertly in connection with the interactive user responses.

7. The multi-modal authentication method of claim 1, wherein the audio features extracted from vocals of the user include at least one of a frequency domain spectrum of the vocals of the user, a pitch of the vocals of the user, power of the vocals of the user, a tone of the vocals of the user, and a cadence of the vocals of the user.

8. The multi-modal authentication method of claim 1, wherein the visual features extracted from an image or video of the user include at least one of position and movement of one or more respective facial features of the user.

9. The multi-modal authentication method of claim 1, further comprising:

computationally determining a lack of correspondence between either or both of (i) the enrollment dataset and the first- or second-type behavioral biometrics and (ii) the time-aligned audio and visual features; and

responsive to the determined lack of correspondence;

presenting the user with supplemental identity validation task and capturing the user'"'"'s response thereto; and

authenticating the identity of the user based on the captured user response to the supplemental identity validation task.

10. The multi-modal authentication method of claim 9, wherein the supplemental identity validation task includes speaking a passphrase.

11. The multi-modal authentication method of claim 9,

wherein the supplemental identity validation task includes typing a passphrase;

and

wherein the identity authenticating includes computationally evaluating correspondence of captured key sequence timings with key sequence timings previously captured for, and previously determined to be, characteristic of the user.

12. The multi-modal authentication method of claim 1, wherein the tracking of the on-screen moving target includes controlling a position of an on-screen aiming mechanism in correspondence with movements of the user'"'"'s head.

13. The multi-modal authentication method of claim 1, wherein the tracking of the on-screen moving target includes tracking of the on-screen moving target based on movements of the user'"'"'s head.

14. The multi-modal authentication method of claim 1, wherein the tracking of the on-screen moving target includes manipulations of the user'"'"'s head or facial presentation.

15. An authentication method, comprising:

capturing, by way of a computing device or an audio interface, a sequence of audio features extracted from vocals of a particular user, wherein the captured sequence of audio features corresponds to a training speech sequence, and wherein the captured sequence of audio features includes one or more of a frequency domain spectrum, a pitch, a power, a tone, and a cadence of the vocals of the particular user;

capturing, by way of the computing device or an image or video interface, a sequence of visual features extracted from an image or video of the particular user, wherein the captured sequence of visual features includes one or more facial movements corresponding to the captured sequence of audio features;

from the captured sequence of corresponding audio and visual features, computationally evaluating at least a portion of the training speech sequence including computational evaluation of the corresponding facial movements and the one or more of the frequency domain spectrum, the pitch, the power, the tone, and the cadence of the vocals of the particular user, thereby determining an audio-visual biometric characteristic of the particular user and suitable to discriminate the training speech sequence spoken by the particular user from the training speech sequence spoken by a statistically significant set of other users;

storing the audio-visual biometric for use in future authentication of the particular user based on an authentication speech sequence spoken by the particular user, wherein the authentication speech sequence does not directly correspond to the training speech sequence, but shares the computationally evaluated portion of the training speech sequence determined to provide the audio-visual biometric; and

based on the stored audio-visual biometric, computationally evaluating portions of speech sequences, in the course of an interactive session, by a user at a computing device or interface and, based on correspondence with the audio-visual biometric, authenticating or confirming authentication of the particular user;

wherein during the interactive session, the user provides an interactive response including tracking of an on-screen moving target using an on-screen aiming mechanism for detecting and tracking facial landmarks employed to improve quality or uniformity of the image or video of the particular user.

16. The authentication method of claim 15, further comprising:

based on the stored audio-visual biometric, generating and supplying a challenge to a user who purports to be the particular user, wherein the challenge solicits speech entry at a computing device or interface and wherein the solicited speech entry includes the portion of the training speech sequence that the computational evaluation of the corresponding facial movements and the one or more of the frequency domain spectrum, the pitch, the power, the tone, and the cadence of the vocals of the particular user determined provide the audio-visual biometric.

17. The authentication method of claim 15, further comprising:

performing, at successive times throughout the interactive session, the computational evaluating of portions of speech sequences for correspondence with the audio-visual biometric.

18. The authentication method of claim 17,

wherein the computational evaluating, at successive times throughout the interactive session, of portions of speech sequences for correspondence with the audio-visual biometric is covert.

19. The authentication method of claim 15, further comprising:

capturing the sequence of corresponding audio and visual features corresponding to the authentication speech sequence, in ordinary course of the interactive session, by a user who purports to be the particular user.

20. The authentication method of claim 15, further comprising:

performing (i) capture of the corresponding sequences of audio and visual features and (ii) the determination of the portion of the training speech sequence determined to provide the audio-visual biometric in a networked computing environment wherein a user enters speech at a client user interface system as part of a session with a remote, network connected service platform; and

communicating, from the client user interface system to the remote, network connected service platform, the facial movements and the corresponding one or more of the frequency domain spectrum, the pitch, the power, the tone, and the cadence of the vocals of the particular user for the portion of the training speech appearing in the captured sequence of corresponding audio and visual features.

21. An authentication method, comprising:

generating and supplying, at a computing device or interface, an on-screen moving target in a first position and an on-screen aiming mechanism for detecting and tracking facial landmarks in image sequences captured with an image sensor, wherein the detecting and tracking facial landmarks is used for aiming at the on-screen moving target in the first position;

capturing, by way of a computing device having or interface coupled to an image sensor directed toward a particular user, motion of visual features including gross user motion or fine user motion including motion of one or more facial landmarks;

based on the captured motion of the visual features, moving the on-screen aiming mechanism correspondingly to, in magnitude and direction, the captured motion of the visual features, wherein the motion of the visual features and corresponding movement of the on-screen aiming mechanism brings the on-screen aiming mechanism into at least a partially overlapping alignment with the on-screen moving target in the first position; and

with the on-screen aiming mechanism in the at least partially overlapping alignment with the on-screen moving target in the first position, capturing by way of the computing device or interface, an image of the particular user in a first position, wherein the image of the particular user in the first position is computationally evaluated and a corresponding first position score is generated, where the generated first position score is used for user recognition and authentication.

22. The authentication method of claim 21,

wherein movement of the on-screen aiming mechanism is in correspondence with the tracking of the one or more of the facial landmarks across a captured visual field.

23. The authentication method of claim 21, further comprising:

based on the captured motion of the visual features, moving the on-screen aiming mechanism into at least a partially overlapping alignment with the on-screen moving target in a second position; and

with the on-screen aiming mechanism in the at least partially overlapping alignment with the on-screen moving target in the second position, capturing, by way of the computing device or interface, an image of the particular user in a second position, wherein the image of the particular user in the second position is computationally evaluated and a corresponding second position score is generated, where the generated second position score is used for user recognition and authentication.

24. The authentication method of claim 23, further comprising:

processing, by way of the computing device or interface, the generated first and second scores;

determining that the processed first and second scores satisfy a threshold confidence level; and

authenticating an identity of the particular user based on the processed first and second scores satisfying the threshold confidence level.

25. The authentication method of claim 23, further comprising:

setting a user identification for the particular user;

capturing, by way of the computing device or interface, an image of the particular user in a first training position, wherein the image of the particular user in the first training position is computationally evaluated and a corresponding first training position score is generated, where the generated first training position score is used for user recognition and authentication;

capturing, by way of the computing device or interface, an image of the particular user in a second training position, wherein the image of the particular user in the second training position is computationally evaluated and a corresponding second training position score is generated, where the generated second training position score is used for user recognition and authentication; and

storing a visual biometric based at least in part on the generated first and second training position scores, wherein the visual biometric corresponds to the user identification for the particular user.