Method for data processing and related products

• US 10,642,820 B2
• Filed: 06/13/2018
• Issued: 05/05/2020
• Est. Priority Date: 06/21/2017
• Status: Active Grant

• Current Assignee: Guangdong OPPO Mobile Telecommunications Corporation Limited
• Original Assignee: Guangdong OPPO Mobile Telecommunications Corporation Limited
• Inventors: Bai, Jian
• Primary Examiner(s): Lewis, Cheryl
• Application Number:
  US16/007,496
• Publication Number:
  US 20180373749A1
• Time in Patent Office: 692 Days
• CPC Class Codes: G06F 16/23 Updating
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06N 5/045 Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence

Abstract

A method for data processing and related products are provided. A terminal device includes at least one processor and a computer readable storage coupled to the at least one processor. The computer readable storage stores at least one computer executable instruction thereon, which when executed by the at least one processor, cause the at least one processor to carry out actions, including: acquiring first user-habit data; determining a similarity value based on the first user-habit data and standard user-habit data; extracting, from the first user-habit data, normal user-habit data in response to the determination that the similarity value is smaller than a first threshold similarity value; updating the standard user-habit data according to the normal user-habit data.

Claims

1. A terminal device, comprising:

at least one processor; and

a computer readable storage, coupled to the at least one processor and storing at least one computer executable instruction thereon, which when executed by the at least one processor, cause the at least one processor to;

acquire first user-habit data;

determine a similarity value based on the first user-habit data and standard user-habit data;

extract, from the first user-habit data, normal user-habit data in response to a determination that the similarity value is smaller than a first threshold similarity value; and

update the standard user-habit data according to the normal user-habit data.

2. The terminal device of claim 1, wherein the instructions to determine the similarity value based on the first user-habit data and standard user-habit data further comprise instructions, which when executed by the at least one processor, cause the at least one processor to:

extract features from the first user-habit data to obtain P parameters, wherein P is an integer greater than 1;

acquire, from the standard user-habit data, target standard user-habit data corresponding to a generation time of the first user-habit data;

extract features from the target standard user-habit data to obtain Q parameters, wherein Q is an integer equal to or greater than P; and

determine the similarity value according to the Q parameters and the P parameters.

3. The terminal device of claim 2, wherein the instructions to determine the similarity value according to the Q parameters and the P parameters further comprise instructions, which when executed by the at least one processor, cause the at least one processor to:

determine variance or covariance between the P parameters and the Q parameters as the similarity value.

4. The terminal device of claim 1, wherein the first user-habit data comprises M data records, and M is an integer greater than 1, wherein

the instructions to extract, from the first user-habit data, the normal user-habit data further comprise instructions, which when executed by the at least one processor, cause the at least one processor to:

analyze the M data records to obtain M parameter sets;

filter the M parameter sets to obtain N parameter sets;

train the N parameter sets with a preset negative-sample classifier to obtain K parameter sets, wherein K is a positive integer; and

determine data records corresponding to the K parameter sets as the normal user-habit data.

5. The terminal device of claim 1, wherein the instructions to update the standard user-habit data according to the normal user-habit data further comprise instructions, which when executed by the at least one processor, cause the at least one processor to:

extract features from the normal user-habit data to obtain a first feature set;

extract features from the standard user-habit data to obtain a second feature set;

determine an adjustment parameter set corresponding to the second feature set according to the first feature set and the second feature set;

adjust control parameters corresponding to a preset positive-sample classifier according to the adjustment parameter set; and

adjust the standard user-habit data through the preset positive-sample classifier adjusted.

6. The terminal device of claim 1, wherein the instructions further comprise instructions, which when executed by the at least one processor, cause the at least one processor to:

count a number of times of occurrence of abnormal events; and

acquire the first user-habit data when the number of times of occurrence of abnormal events is greater than a second threshold similarity value.

7. The terminal device of claim 1, wherein the at least one processor comprises an application processor and an artificial intelligence (AI) module, wherein the AI module is integrated with or separated from the application processor and configured to perform at least one of:

acquiring the first user-habit data;

determining the similarity value based on the first user-habit data and standard user-habit data;

extracting, from the first user-habit data, the normal user-habit data in response to the determination that the similarity value is smaller than a first threshold similarity value; and

updating the standard user-habit data according to the normal user-habit data.

8. The terminal device of claim 1, wherein the standard user-habit data is obtained by training usage patterns of a user for at least one application over a period of time by the terminal device.

9. The terminal device of claim 1, wherein the first user-habit data is obtained from usage records of the user, and one usage record is generated each time a user operates the terminal device.

10. A method for data processing, comprising:

acquiring first user-habit data;

determining a similarity value based on the first user-habit data and pre-stored standard user-habit data;

extracting, from the first user-habit data, normal user-habit data in response to a determination that the similarity value is smaller than a first threshold similarity value; and

updating the pre-stored standard user-habit data according to the normal user-habit data.

11. The method of claim 10, wherein determining the similarity value based on the first user-habit data and the pre-stored standard user-habit data comprises:

extracting features from the first user-habit data to obtain P parameters, wherein P is an integer greater than 1;

acquiring, from the pre-stored standard user-habit data, target standard user-habit data corresponding to a generation time of the first user-habit data; and

extracting features from the target standard user-habit data to obtain Q parameters and determining the similarity value according to the Q parameters and the P parameters, wherein Q is an integer equal to or greater than P.

12. The method of claim 11, wherein determining the similarity value according to the Q parameters and the P parameters comprises:

determining variance or covariance between the P parameters and the Q parameters as the similarity value.

13. The method of claim 10, wherein the first user-habit data comprises M data records and M is an integer greater than 1, and the extracting, from the first user-habit data, the normal user-habit data comprises:

analyzing the M data records to obtain M parameter sets and filtering the M parameter sets to obtain N parameter sets; and

training the N parameter sets with a negative-sample classifier to obtain K parameter sets, wherein K is a positive integer, and determining data records corresponding to the K parameter sets as the normal user-habit data.

14. The method of claim 13, wherein the negative-sample classifier is constructed via negative samples obtained from abnormal user-habit data.

15. The method of claim 10, wherein the updating the pre-stored standard user-habit data according to the normal user-habit data comprises:

extracting features from the normal user-habit data to obtain a first feature set;

extracting features from the pre-stored standard user-habit data to obtain a second feature set;

determining an adjustment parameter set corresponding to the second feature set according to the first feature set and the second feature set;

adjusting control parameters corresponding to a positive-sample classifier according to the adjustment parameter set; and

adjusting the pre-stored standard user-habit data through a preset positive-sample classifier adjusted.

16. The method of claim 15, wherein the positive-sample classifier is constructed via positive samples obtained from normal user-habit data.

17. A non-transitory computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to:

acquire first user-habit data;

determine a similarity value based on the first user-habit data and pre-stored standard user-habit data;

extract, from the first user-habit data, normal user-habit data in response to a determination that the similarity value is smaller than a threshold similarity value; and

update the pre-stored standard user-habit data according to the normal user-habit data.

18. The non-transitory computer-readable storage medium of claim 17, wherein the computer program executed by the processor to determine the similarity value based on the first user-habit data and the pre-stored standard user-habit data further comprise computer programs to:

extract features from the first user-habit data to obtain P parameters, wherein P is an integer greater than 1;

acquire, from the pre-stored standard user-habit data, target standard user-habit data corresponding to a generation time of the first user-habit data;

extract features from the target standard user-habit data to obtain Q parameters, wherein Q is an integer equal to or greater than P; and

determine the similarity value according to the Q parameters and the P parameters.

19. The non-transitory computer-readable storage medium of claim 17, wherein the computer program executed by the processor to extract, from the first user-habit data, the normal user-habit data further comprise computer programs to:

analyze M data records contained in the first user-habit data to obtain M parameter sets, wherein M is an integer greater than 1;

filter the M parameter sets to obtain N parameter sets;

train the N parameter sets with a preset negative-sample classifier to obtain K parameter sets, wherein K is a positive integer; and

determine data records corresponding to the K parameter sets as the normal user-habit data.

20. The non-transitory computer-readable storage medium of claim 17, wherein the computer program executed by the processor to update the pre-stored standard user-habit data according to the normal user-habit data further comprise computer programs to:

extract features from the normal user-habit data to obtain a first feature set;

extract features from the pre-stored standard user-habit data to obtain a second feature set;

determine an adjustment parameter set corresponding to the second feature set according to the first feature set and the second feature set;

adjust control parameters corresponding to a preset positive-sample classifier according to the adjustment parameter set; and

adjust the pre-stored standard user-habit data through the preset positive-sample classifier adjusted.