Method and device for determining sample labels

Abstract

The invention provides a method and a device for determining sample labels. The method for determining the sample labels comprises the following steps of extracting on-line feedback data to acquire initial model training data; extracting scene features from the initial model training data; adding the scene features into the initial model training data to acquire sample data; calculating the sample quality score of each group of sample data; labeling each group of sample data according to the corresponding sample quality score of the corresponding group of sample data; learning the labeled sample data to generate a model; and predicting labels of the sample data by using the model. In a multi-example learning problem, labor is reduced when the sample is labeled, a large amount of labor is saved, and the labeling efficiency can be improved.

Description

technical field [0001] The present invention relates to the technical field of the Internet, in particular to a method and device for determining sample labeling. Background technique [0002] In multi-instance learning, a sample is a bag composed of multiple examples. Multi-instance learning is to label samples at the granularity of the bag, and each sample contains several examples, which are not labeled. If a sample is marked as a positive example, there is at least one positive example in the sample; otherwise, if a sample is marked as a negative example, all examples in the sample are negative examples. The purpose of multi-instance learning is to make judgments on new samples as accurately as possible by learning from these labeled samples. [0003] In the multi-instance learning problem, the existing technology mainly screens and labels the samples manually. This labeling method has high labor costs and low labeling efficiency. Contents of the invention [0004] T...

AI Technical Summary

Problems solved by technology

[0003] In the multi-instance learning problem, the existing technology mainly screens and labels samples manually. This labeling method has high labor costs and low labeling efficiency.

Images