Hyper-parameter determination method and device, computer equipment and storage medium

Abstract

The embodiment of the invention discloses a hyper-parameter determination method and device, computer equipment and a storage medium. According to the embodiment of the invention, the method comprisesthe steps: obtaining a hyper-parameter group of at least one to-be-trained model, wherein the to-be-trained model belongs to a target model type, the hyper-parameter group comprises hyper-parametersof at least one dimension, and the hyper-parameters of each dimension indicate model structure features of the to-be-trained model under each dimension; distributing the target hyper-parameter group to a target hyper-parameter evaluation node in the distributed cluster; through the target hyper-parameter evaluation node, adopting the target hyper-parameter set to perform model training on a targetto-be-trained model, wherein the target to-be-trained model is a to-be-trained model corresponding to the target hyper-parameter set; performing model evaluation on the trained model to obtain a hyper-parameter evaluation result of each hyper-parameter group; based on the hyper-parameter evaluation result, selecting an optimal hyper-parameter group of the target model type from the hyper-parameter groups. According to the scheme, the hyper-parameters can be efficiently determined by improving the hyper-parameter tuning efficiency.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to a hyperparameter determination method, device, computer equipment and storage medium. Background technique [0002] In the context of machine learning, a hyperparameter is a parameter whose value is set before starting the learning process, rather than the parameter data obtained through training. Usually, it is necessary to optimize the hyperparameters and select a set of optimal hyperparameters for the learning machine to improve the performance and effect of learning. [0003] Many machine learning models need to pre-set hyperparameter values ​​before training, and hyperparameter values ​​play an important role in improving the learning performance and effect of the model. Especially when the number of hyperparameters is large, how to determine the value of each hyperparameter poses a big challenge to the modeler. The inventors of the present application have fou...

AI Technical Summary

Problems solved by technology

Especially when the number of hyperparameters is large, how to determine the value of each hyperparameter poses a big challenge to the modeler

The inventors of the present application found that at present, empirical methods are usually used to determine directly, which makes the tuning efficiency of hyperparameters very low, and the method of determining hyperparameters is also very limited, which urgently needs to be improved

Images