A model training method and device

Abstract

The embodiment of the present invention discloses a model training method and device, wherein the method includes: for each negative sample data, according to the difference between the negative sample data and P positive sample data and the negative sample data and Q negative samples The difference value of the data to determine whether the negative sample data is unreliable negative sample data; use P positive sample data and Q negative sample data except the credible negative sample data for model training, and get two classification model. In the embodiment of the present invention, the credible negative sample data used to train the model is obtained by determining the untrustworthy negative sample data among the Q negative sample data, compared with the negative sample data obtained based on random screening or manual screening in the prior art In other words, the credible negative sample data in the embodiment of the present invention is more accurate, so that the binary classification model trained based on the credible negative sample data is more reasonable, and the prediction result is more in line with the actual situation.

Description

technical field [0001] The invention relates to the field of data processing, in particular to a model training method and device. Background technique [0002] In the Internet marketing mode, marketers usually hope to be able to screen out users (called target users) who are interested in the products they market from among multiple users, so that they can place advertisements to target users in a targeted manner to achieve product marketing. One of the most commonly used methods for determining target users is: use positive sample data and negative sample data to train the model to obtain a binary classification model, and use the trained binary classification model to predict whether a certain user is a target user. Among them, the positive sample data can be the data of existing users who are interested in the product, for example, the data of users who have purchased the product included in the marketer's customer system; the negative sample data can be filtered from th...

AI Technical Summary

Problems solved by technology

Among them, the first method may not be able to guarantee the quality of the negative sample data due to the use of random screening to determine the negative sample data. Therefore, if the negative sample data is used for model training, the effect of the trained model may be poor.

The second method uses manual screening, the cost is relatively high, the screening time is long, and the efficiency is not high; and the negative sample data manually screened cannot meet the needs of different marketing products, resulting in poor flexibility for model training. not effectively

Images