Processing machine learning attributes

Patent metadata

Field	Value
Publication number	US-11301765-B2
Application number	US-201615296520-A
Country	US
Kind code	B2
Filing date	Oct 18, 2016
Priority date	Oct 18, 2016
Publication date	Apr 12, 2022
Grant date	Apr 12, 2022

Abstract

Systems and methods for processing machine learning attributes are disclosed. An example method includes: identifying a user transaction associated with a set of transaction attributes and a first transaction status; selecting, based on a risk evaluation model, a first plurality of transaction attributes from the set of transaction attributes; modifying a first value of a first transaction attribute in the first plurality of transaction attributes to produce a first modified plurality of transaction attributes; determining, based on the risk evaluation model, that the first modified plurality of transaction attributes identify a second transaction status different from the first transaction status; and in response to the determining, identifying the first transaction attribute as a risk attribute associated with the user transaction.

First claim

What is claimed is:

1. A system, comprising:
a non-transitory memory; and
one or more hardware processors coupled to the non-transitory memory and configured to execute instructions to cause the system to perform operations comprising:
receiving, from a user device via a merchant device, a request to perform an electronic transaction between a user of the user device and a merchant associated with the merchant device, wherein the request comprises a plurality of attribute values corresponding to a plurality of transaction attributes and associated with the electronic transaction;
determining, using a risk evaluation model, that the request is associated with a first risk above a predefined amount;
denying the request based on the first risk being above the predefined amount;
determining, from the plurality of transaction attributes and based on analyzing the risk evaluation model, a set of machine learning attributes used by the risk evaluation model for evaluating electronic transactions;
generating a plurality of modified transaction requests by iteratively selecting different subsets of the set of machine learning attributes and modifying attribute values corresponding to the different subsets of the set of machine learning attributes;
determining, using the risk evaluation model, at least one machine learning attribute from the set of machine learning attributes that contributes to the first risk being above the predefined amount based on the plurality of modified transaction requests;
presenting an indication that the request is denied and the at least one machine learning attribute that contributed to the first risk associated with the request being above the predefined amount;
prompting, on a user interface of the user device, the user for additional information based on the at least one machine learning attribute; and
processing the electronic transaction based on the additional information.

2. The system of claim 1 , wherein the operations further comprise:
determining, using the risk evaluation model, a plurality of risks associated with the plurality of modified transaction requests;
identifying, from the plurality of modified transaction requests, a first subset of modified transaction requests associated with first risks below the predefined amount and a second subset of modified transaction requests associated with second risks above the predefined amount; and
determining a difference in attribute values between the first subset of modified transaction requests and the second subset of modified transaction requests.

3. The system of claim 1 , wherein the operations further comprise:
receiving, from the user device, an inquiry for a reason for the denying the request; and
providing the at least one risk transaction attribute on the user interface of the user device in response to receiving the inquiry.

4. The system of claim 3 , wherein the operations further comprise:
withholding the set of machine learning attributes, less the at least one machine learning attribute, from being provided on the user device.

5. The system of claim 1 , wherein the selecting and the modifying are performed iteratively for a predefined number of times to generate the plurality of modified transaction requests.

6. The system of claim 5 , wherein the operations further comprise:
receiving an inquiry from the user regarding the denying of the request;
determining that the inquiry is received via a particular communication channel; and
determining a predefined number of times for iteratively performing the selecting and the modifying based on the determining that the inquiry is received via the particular communication channel.

7. The system of claim 1 , wherein the operations further comprise:
identifying training data attributes used to train the risk evaluation model, and wherein the different subsets of the set of machine learning attributes are selected based on the training data attributes.

8. The system of claim 7 , wherein the different subsets of the set of machine learning attributes match the training data attributes.

9. A method, comprising:
receiving, over a network connection from a user device via a merchant device, a request to perform an online transaction between a user of the user device and a merchant associated with the merchant device, wherein the request comprises a plurality of attribute values corresponding to a plurality of transaction attributes and associated with the online transaction;
determining, by one or more hardware processors using a risk evaluation model, that the request is associated with a first risk above a predefined amount based on the plurality of attribute values;
denying the request based on the first risk being above the predefined amount;
determining, from the plurality of transaction attributes and based on analyzing the risk evaluation model, a set of transaction attributes used by the risk evaluation model for evaluating electronic transactions;
generating, by the one or more hardware processors, a plurality of modified transaction requests by iteratively selecting different subsets of the set of transaction attributes and modifying attribute values corresponding to the different subsets of the set of transaction attributes;
determining, by the one or more hardware processors using the risk evaluation model, at least one risk transaction attribute from the set of transaction attributes that contributes to the first risk being above the risk threshold based on the plurality of modified transaction requests;
providing, over the network connection to the merchant device, an indication that the request is denied and the at least one risk transaction attribute that contributed to the first risk associated with the request being above the predefined amount;
prompting, on the user device, for additional information based on the at least one risk transaction attribute; and
processing the online transaction based on the additional information.

10. The method of claim 9 , further comprising:
determining, using the risk evaluation model, a plurality of risks associated with the plurality of modified transaction requests;
identifying, from the plurality of modified transaction requests, a first subset of modified transaction requests associated with first risks below the predefined amount and a second subset of modified transaction requests associated with second risks above the predefined amount; and
determining a difference in attribute values between the first subset of modified transaction requests and the second subset of modified transaction requests.

11. The method of claim 10 , wherein the first subset of modified transaction requests and the second subset of modified transaction requests share no overlapping modified attribute values.

12. The method of claim 9 , further comprising:
withholding the set of transaction attributes, except the at least one risk transaction attribute, from being provided on the user device.

13. The method of claim 9 , wherein the selecting and the modifying are performed iteratively for a number of times to generate the plurality of modified transaction requests.

14. The method of claim 13 , further comprising:
determining the number of times for iteratively performing the selecting and the modifying based on a communication channel through which the request is received.

15. The method of claim 9 , further comprising:
identifying training data attributes used to train the risk evaluation model, wherein the different subsets of the set transaction attributes are selected based on the training data attri

Assignees

• Paypal Inc

Inventors

• Lin Ying
• Yin Huagang
• Zhang Jiaqi

Classifications

• G06N5/045 Primary

  Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence · CPC title

• G06Q20/4016

  involving fraud or risk level assessment in transaction processing · CPC title