A prediction model training method and device for new scenarios

Abstract

Disclosed are a method and device for training a prediction model for a new scenario. The method for training a prediction model for a new scenario comprises: acquiring a set of models to be migrated; selecting at least one model from the set of models to predict annotations for untagged samples in a new scenario; obtaining an initial training sample set in the new scenario; using the selected model to add predicted tags to untagged samples in the initial training sample set; and using the initial training sample set having been added with the predicted tags to update the model on the basis of a supervised learning algorithm so as to obtain a model applicable to the new scenario.

Description

technical field [0001] The embodiments of this specification relate to the field of Internet application technologies, and in particular to a method and device for training a prediction model for new scenarios. Background technique [0002] In the era of big data, models can be trained through machine learning based on accumulated sample data to achieve the required decision-making functions. For example, in the financial risk control scenario, a large amount of transaction data can be used as sample data to train the risk control model through machine learning, so that the trained risk control model can be used to automatically make risk decisions for new transactions. [0003] However, in some scenarios, it often takes a long time to accumulate sample data and train the model to implement the deployment of the machine learning model. For example, the data accumulation and training of the risk control model generally takes more than half a year. In this regard, one solutio...

AI Technical Summary

Problems solved by technology

[0003] However, in some scenarios, it often takes a long time to accumulate sample data and train the model to realize the deployment of the machine learning model. For example, the data accumulation and training of the risk control model generally takes more than half a year

In this regard, one solution is to deploy a historical model in the new scene. The historical model is trained based on the historical data in other scenes. However, due to the differences between the sample data of each scene, the historical model is applied in the new scene. is usually less effective

Images