Intelligent model training method under industrial Internet of Things

Abstract

The invention discloses an intelligent model training method under the industrial Internet of Things, and the method comprises the specific steps: selecting a training set of a user: enabling the userto import the training set needing to be trained, and extracting the features of a current training set; searching a similarity test set: according to the characteristics of the current training set,finding a historical test set similar to the current training set in the system; conducting model selection: according to a certain historical test set selected by the user, finding a model test statistical result using the historical test set at the moment in a matched mode; importing training parameters: importing the training parameters when the model is trained at the moment according to a certain model selected by the user; carrying out intelligent training: directly importing an algorithm of the model and automatically starting training, or manually starting training after the user adjusts some parameters; and outputting the model: obtaining a model corresponding to the current user training set after the training is finished. According to the model intelligent training method, starting from a similarity test set in the system, the labor cost and the time cost of algorithm engineers are greatly reduced, and the production efficiency is improved to the maximum extent.

Description

technical field [0001] The invention relates to the technical field of training methods for the Industrial Internet of Things, in particular to a model intelligent training method under the Industrial Internet of Things. Background technique [0002] Artificial Intelligence (AI) is a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. [0003] The training process of the traditional AI model is: Step 1, sample labeling (labeling the provided sample data, such as manually identifying and labeling a picture, whether there are defects, and if so, mark the defects); The marked samples are divided into training set and test set (equivalent to artificially obtaining a standard answer); step 3, continuously adjust the parameters, and find excellent performance from the train...

AI Technical Summary

Problems solved by technology

Obviously, the traditional AI model training process is cumbersome, time-consuming to adjust parameters, and the model training results cannot be reused

Images