Machine learning model adjustment method and device based on ensemble learning

Abstract

The invention relates to a machine learning model adjustment method and a device based on ensemble learning. The method comprises the following steps: constructing an initial classification model; the initial classification model is a machine learning model; performing k rounds of iterative adjustment on the initial classification model by adopting a BF method to reduce the deviation of the initial classification model; a BBF method is adopted to adjust the initial classification model for multiple times, and the variance of the initial classification model is reduced; carrying out average output on multiple times of adjustment, and finally obtaining an adjusted classification model; the device is used for implementing the method. The invention has the advantages that the performance of a machine learning model is improved, classification can be better carried out on the image target classification problem with small data size, the problem that a single model is prone to overfitting on data sets is solved, and therefore the target classification precision is improved.

Description

technical field [0001] The invention relates to the field of machine learning, in particular to a method and device for adjusting a machine learning model based on integrated learning. Background technique [0002] With the rapid development of key technologies such as the Internet, cloud computing, and big data, artificial intelligence technology is profoundly changing people's lives and driving changes in all aspects of society, such as finance, entertainment, education, medical care, and social networking. Machine learning is an important part of artificial intelligence, which aims to learn the potential relationship between data from a large amount of data, and has great potential for business mining. Many of these algorithms have permeated various branches of artificial intelligence, including autonomous driving, recommendation systems, face recognition, natural language processing, and more. [0003] Integrated learning plays an important role in machine learning and ...

AI Technical Summary

Problems solved by technology

However, for a non-guidable model such as kernel SVM, it cannot be adjusted using gradient descent, and in data scenarios with a small amount of data, kernel SVM is a strong benchmark, and most methods are difficult to compare with kernel SVM.

However, only using a single learner for prediction, the output is still more susceptible to the influence of noise points, resulting in poor generalization ability of the model

For RLR and RR, although gradient descent can be used to adjust them, their models after training are already globally optimal. Without changing the loss function, a better model cannot be obtained through adjustment. And after adjustment, only one model will be used for prediction, so their output results are also easily affected by noise points

When training a model on a medical disease classification dataset, it is difficult to select appropriate model parameters due to the small amount of data and noise in the dataset, so it is easy to overfit when only using a single model, resulting in poor classification effect. it is good

Images