Granted Patent B2

US 12,341,913 · App. 18/165,281 · Granted Jun 24, 2025

Inventors: Samuel Rapowitz (Roswell, GA); Zviad Aznaurashvili (Reston, VA); Viraj Chaudhary (Katy, TX)

Assignee: CAPITAL ONE SERVICES, LLC

H04L9/3271G06N20/00G06Q20/389

Abstract

Claims (57)

1. An authentication method comprising:

receiving, from a user device in communication with an authentication server, an authentication request associated with a user in possession of the user device;

requesting, from a transaction history database, a transaction history associated with the user, the transaction history comprising a first purchase of a first product and a second purchase of a second product;

selecting, using a machine learning model, the first product and the second product to authenticate the user;

identifying, based on the transaction history associated with the user, a third product that the user did not purchase;

generating, based on stock keeping unit (SKU) data, images of the first product, the second product, and the third product;

generating, by the machine learning model, an authentication quiz comprising:

the images of the first product, the second product, and the third product; and

requests to the user to confirm that the user purchased one or more of the first product, the second product, and the third product;

receiving, from the user device, responses from the user confirming that the user purchased the first product and the second product and denying that the user purchased the third product; and

transmitting instructions to the user device to authenticate the user.

2. The method of claim 1 , wherein the machine learning model comprises a machine learning fraud prediction model.

3. The method of claim 1 , wherein the first product and the second product are selected based on a predetermined difficulty associated with the authentication request.

4. The method of claim 3 , wherein the predetermined difficulty is calculated based on a recency of the first purchase and the second purchase.

5. The method of claim 3 , wherein the predetermined difficulty is calculated based on a likelihood of the user repeating the first purchase and the second purchase.

6. The method of claim 3 , wherein the predetermined difficulty is calculated based on a confidence threshold set by the authentication server.

7. An authentication method comprising:

receiving, from a user device in communication with an authentication server, an authentication request associated with a user in possession of the user device;

requesting, from a transaction history database, a transaction history associated with the user, the transaction history comprising a first purchase of a first product;

selecting, using a machine learning model, the first product to authenticate the user;

generating, by the machine learning model, an authentication quiz comprising:

an image of the first product, the image being obtained based on stock keeping unit (SKU) data of the first product; and

transmitting instructions to the user device to authenticate the user based on the first response.

8. The method of claim 7 , wherein the transaction history further comprises a second purchase of a second product, and wherein the method further comprises:

selecting, by the machine learning model, the second product to authenticate the user;

updating, by the machine learning model, the authentication quiz to include:

an image of the second product, the image being obtained based on SKU data of the second product; and

a request to the user to confirm that the user purchased the second product; and

receiving, from the user device, a second response from the user confirming that the user purchased the second product, wherein transmitting instructions to the user device to authenticate the user is further based on the second response.

9. The method of claim 8 , further comprising:

responsive to receiving the second response, selecting, by the machine learning model, a third product to authenticate the user, the third product being not purchased by the user;

updating, by the machine learning model, the authentication quiz to include:

an image of the third product, the image being obtained based on SKU data of the third product; and

a request to the user to confirm that the user purchased the third product; and

receiving, from the user device, a third response from the user denying that the user purchased the third product.

10. The method of claim 8 , wherein the first purchase is associated with a first merchant and the second purchase is associated with a second merchant, and the images of the first product and the second product are obtained by requesting the first merchant and the second merchant, respectively, with the SKU data of the first product and the second product.

11. The method of claim 7 , wherein the machine learning model comprises a machine learning fraud prediction model.

12. The method of claim 8 , wherein the first product and the second product are selected based on a predetermined difficulty associated with the authentication request.

13. The method of claim 12 , wherein the predetermined difficulty is calculated based on a likelihood of the user repeating the first purchase and the second purchase.

14. The method of claim 12 , wherein the predetermined difficulty is calculated based on a confidence threshold set by the authentication server.

15. The method of claim 12 , wherein the predetermined difficulty is calculated based on a recency of the first purchase and the second purchase.

16. The method of claim 15 , wherein the transaction history associated with the user is requested based to include a number of purchases based on the recency and the predetermined difficulty.

17. An authentication method comprising:

receiving, from a user device in communication with an authentication server, an authentication request associated with a user in possession of the user device;

requesting, from a transaction history database, a transaction history associated with the user, the transaction history comprising a first purchase of a first product and a second purchase of a second product;

selecting, using a machine learning model, the first product and the second product to authenticate the user;

determining a first merchant associated with the first purchase and a second merchant associated with the second purchase;

generating, based on stock keeping unit (SKU) data, images of the first product and the second product;

generating, by the machine learning model, an authentication quiz comprising:

the images of the first product and the second product;

an indication of the first merchant and the second merchant; and

a request to the user to match the first product and the second product to the corresponding merchant of the first merchant and the second merchant where the first purchase and the second purchase occurred;

receiving, from the user device, responses from the user matching the first product to the first merchant and the second product to the second merchant; and

transmitting instructions to the user device to authenticate the user.

18. The method of claim 17 , wherein the first purchase is associated with a first merchant and the second purchase is associated with a second merchant, and the images of the first product and the second product are obtained by requesting the first merchant and the second merchant, respectively, with the SKU data of the first product and the second product.

19. The method of claim 17 , wherein the machine learning model comprises a machine learning fraud prediction model.

20. The method of claim 17 , wherein the first product and the second product are selected based on a predetermined difficulty associated with the authentication request.

Assignments (1)

Continuity (1)

References Cited (11)