Pre-labeling method based on hierarchical transfer learning and related equipment thereof

Abstract

The embodiment of the invention belongs to the field of artificial intelligence, and relates to a pre-labeling method based on hierarchical transfer learning and related equipment thereof, and the method comprises the steps: carrying out the clustering of a plurality of different scenes received in advance based on a preset clustering algorithm, and obtaining a clustering result; determining a first type of scenes and a second type of scenes according to the clustering result, the first type of scenes including the first scene, and the data volume of the annotation data in the first scene being greater than the data volume of the annotation data of any scene in the second type of scenes; performing transfer learning on the first type of scenes based on a preset identification model to obtain pre-annotation data and a transfer model of each scene in the first type of scenes; and based on a migration model, carrying out migration learning on the second type of scenes to obtain pre-annotation data of each scene in the second type of scenes. The pre-annotation data of each scene can be stored in the block chain. According to the method and the device, better pre-annotation data in different scenes can be quickly obtained.

Description

technical field [0001] This application relates to the field of artificial intelligence technology, and in particular to a pre-labeling method based on hierarchical transfer learning and related equipment. Background technique [0002] With the development of random technology, intelligent recognition technology has been widely used. Many data no longer need to be labeled by personnel, but can be directly pre-labeled through the model. The personnel only need to correct the pre-labeled results, effectively reducing the Labor costs and labeling time. [0003] However, before the model is actually put into use and pre-labeled, a large amount of labeled data is required to train the model, and the effect of model training directly affects the result of pre-labeling. However, when training models for different scenarios, personnel are still required to label a large amount of data in each scenario separately, resulting in high labor costs and slow labeling speed, and it is impo...

AI Technical Summary

Problems solved by technology

[0003] However, before the model is actually put into use and pre-labeled, a large amount of labeled data is required to train the model, and the effect of model training directly affects the result of pre-labeling

However, when training models for different scenarios, personnel are still required to label a large amount of data in each scenario separately, resulting in high labor costs and slow labeling speed, and it is impossible to quickly obtain a large amount of data in different scenarios in a short period of time. better labeled data

[0004] The current method is to directly migrate the trained model in the old scene to the new scene for pre-labeling, and then correct the pre-labeling results by personnel. However, due to the large differences in different scenes, the pre-labeling results of the model are poor. , or even not available at all

Images