Generating a training model based on feedback

Patent metadata

Field	Value
Publication number	US-10346746-B2
Application number	US-201314055387-A
Country	US
Kind code	B2
Filing date	Oct 16, 2013
Priority date	Jul 30, 2012
Publication date	Jul 9, 2019
Grant date	Jul 9, 2019

Abstract

A method and apparatus for generating a training model based on feedback are provided. The method for generating a training model based on feedback, includes calculating an eigenvector of a sample among a plurality of samples; obtaining scores granted by a user for one or more of the plurality of samples in a round, obtaining scores granted by the user for a first number of samples; obtaining scores granted by the user for a second number of samples in response to detecting, based on the eigenvector, an inconsistency between the scores granted by the user for the first number of samples; and generating a training model based on the scores granted by the user for the first and second numbers of samples. A corresponding apparatus is also provided.

First claim

The invention claimed is:

1. An apparatus for generating a training model based on feedback in a rating system architecture configured to learn rating standards adopted by at least one user, the apparatus comprising:
a processor configured to:
calculate an eigenvector of an image data sample among the plurality of image data samples stored in a database;
obtain scores granted by the at least one user for at least one of the plurality of image data samples in at least one round by:
obtaining scores granted by the at least one user for a first number of image data samples; and
obtaining scores granted by the at least one user for a second number of image data samples in response to having detected, based on the eigenvector, an inconsistency in the scores granted by the at least one user for the first number of image data samples; and

generate the training model, stored in non-transitory computer readable storage media, to remove noise caused by user errors during rating of image data samples by modifying the rating system architecture to employ representative samples derived from and being a subset of the plurality of image data samples to be rated by a user, based on the scores granted by the at least one user for the first and second numbers of image data samples, the training model to rate and/or condition data for other image data samples.

2. The apparatus according to claim 1 , wherein the processor is further configured to:
group the plurality of samples into at least one group based on the eigenvector.

3. The apparatus according to claim 2 , wherein the processor is further configured to:
obtain a score granted by the at least one user for at least one sample in each group of the at least one group.

4. The apparatus according to claim 1 , wherein the processor is further configured to:
for one user among the at least one user, provide at least two samples;
obtain original scores granted by the user for the at least two samples; and
normalize the original scores to obtain scores granted by the user for the at least two samples.

5. The apparatus according to claim 1 , wherein the processor is further configured to:
if a difference between scores granted by a first user among the at least one user for similar samples exceeds a first threshold, provide more similar samples to the first user to obtain scores granted by the first user for the more similar samples.

6. The apparatus according to claim 1 , wherein the processor is further configured to:
if a difference between scores granted by similar users for a first sample of the plurality of samples exceeds a second threshold, provide the first sample once again to the similar users respectively to obtain scores granted by the similar users for the first sample.

7. The apparatus according to claim 1 , wherein the processor is further configured to:
if a gap between a difference between original scores granted by one user among the at least one user for two samples and a difference between normalized scores exceeds a third threshold, substitute one of the two samples with another sample of the plurality of samples.

8. The apparatus according to claim 1 , wherein the processor is further configured to:
fill in a scoring matrix the scores granted by the at least one user for the first and second numbers of samples; and
collaboratively filter the scoring matrix to generate the training model.

Assignees

• IBM

Inventors

• Cao Liangliang
• Duan Ning
• Lin Qian
• Wang Chen
• Yan Junchi
• Zhang Xin

Classifications

• G06N99/005

  Physics · mapped topic

• G06N5/02 Primary

  Knowledge representation; Symbolic representation · CPC title

• G06N20/00 Primary

  Machine learning · CPC title