Hyper-parameter threshold range determination method and device, storage medium and electronic equipment

Abstract

The invention relates to a hyper-parameter threshold range determination method and device, a storage medium and electronic equipment. The method comprises the steps of obtaining a target hyper-parameter value used in each training process as a first target hyper-parameter value and obtaining an evaluation index value of a model obtained after each training as a first evaluation index value in a process of carrying out multiple training on the model based on a first training data set; and determining a threshold range of the target hyper-parameter according to the first target hyper-parametervalue and the first evaluation index value. Due to the fact that the determination of the threshold range of the target hyper-parameter refers to the corresponding evaluation index, the appropriate target hyper-parameter threshold range can be selected according to different requirements of users. Thus, hyper-parameter tuning can be carried out subsequently according to the target hyper-parameterthreshold range, and the range of repeated tuning attempts is narrowed, so that the efficiency of hyper-parameter tuning is improved, the labor cost is reduced, and the efficiency of artificial intelligence model training is improved.

Description

technical field [0001] The present disclosure relates to the field of artificial intelligence, in particular, to a method, device, storage medium and electronic equipment for determining a hyperparameter threshold range. Background technique [0002] Artificial intelligence has developed rapidly in the past 30 years, and has been widely used in many disciplines and achieved fruitful results. However, at this stage, artificial intelligence model training is still time-consuming and labor-intensive work, which often requires experts to iteratively tune iteratively based on their own experience. Most of the work involves repeated tuning of the hyperparameters of the artificial intelligence model training algorithm try. In order to improve the efficiency of artificial intelligence model training and reduce labor costs, researchers have proposed methods such as grid search (GridSearch) and Bayesian optimization to achieve automatic tuning of hyperparameters. [0003] However, m...

AI Technical Summary

Problems solved by technology

[0003] However, most of the methods proposed so far still rely on the initial value input by the user. For example, the GridSearch method requires the user to manually input a series of hyperparameter arrays, and the Bayesian optimization method also needs to randomly generate a series of sampling points, that is, the initial hyperparameters The threshold range still depends on personal experience to set

In addition, the above-mentioned Grid Search method can only verify the effect of hyperparameter values ​​within the input range, while Bayesian optimization methods often focus on parameter tuning within the surrounding range of the initial sampling point, and they cannot meet the different needs of users ( For example, model accuracy, training efficiency) set an appropriate threshold search method

Images