Multi-classifier online adjustment method, device, storage medium and electronic equipment

Abstract

The multi-classifier online adjustment method and device provided by this application, after using the multi-classifier to classify the current online data to be processed, does not directly use the classified data to adjust the multi-classifier, but the multi-classifier The online data is cached accordingly until the cached data of each category meets the preset quantity condition (the difference between the quantity of any two categories of data does not exceed the first predetermined threshold), and then the data of each category that meets the preset quantity condition is used to adjust the number Classifier. The solution of this application caches the classified data of each category, and uses the cached data of various categories that meet the preset quantity conditions to adjust the multi-classifier, at least to a certain extent, it can ensure that the data of each category used in the adjustment of the classifier model is in the The distribution of quantity is balanced, thus, it can effectively reduce the performance fluctuation of the classification model in the multi-classification problem, and then overcome the problem of inaccurate classification that may exist in the classification model.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to an online adjustment method, device, storage medium and electronic equipment of a multi-classifier. Background technique [0002] In the field of machine learning, the classification method is a method of inputting the classification model of the data training data that has been marked into the corresponding classification algorithm, and then using the trained classification model to predict the new data category. Among them, the training methods of the classification model include offline training and online training. [0003] At present, the online classification methods corresponding to online training include online gradient descent (OGD, Online Gradient Descent) and stochastic gradient descent (SGD, Stochastic Gradient Descent), etc. However, these online methods have the disadvantage of being difficult to generate sparse solutions, and then predict In the process o...

AI Technical Summary

Problems solved by technology

[0003] At present, the online classification methods corresponding to online training include online gradient descent (OGD, Online Gradient Descent) and stochastic gradient descent (SGD, Stochastic Gradient Descent), etc. However, these online methods have the disadvantage of being difficult to generate sparse solutions, and then predict In the process of classifying new data, there will be problems of high memory usage and time complexity. In view of this, FTRL (Follow The Regularized Leader Proximal) for stochastic gradient descent is proposed in this field to solve the problems existing in the stochastic gradient descent method.

However, in practical applications, the multi-classification model based on this method (ie FTRL) tends to have large performance fluctuations, which may lead to inaccurate classification of the multi-classification model.

Images