集成学习方法研究 |
点此下载全文 |
引用本文:周钢?覮,郭福亮.集成学习方法研究[J].计算技术与自动化,2018,(4):148-153 |
摘要点击次数: 1055 |
全文下载次数: 0 |
|
|
中文摘要:集成学习是当前数据挖掘、机器学习中提升预测精度的重要方法。在介绍集成学习概念、评价标准的基础上,将集成学习划分为基分类器的构建和集成两个阶段,从偏差-方差分解角度,分析集成学习的预测精度主要是通过控制集成模型复杂度和各基分类器差异度实现,研究讨论了集成学习的模型构建阶段的经典算法Bagging、Boosting等,同时分析研究了分类结果集成的普通投票和Stacking方法。 |
中文关键词:集成学习 偏差-方差分解 Bagging Boosting Stacking |
|
Research on Ensemble Learning |
|
|
Abstract:Ensemble learning is considered as an important method to improve the accuracy of data mining and machine learning. Based on introducing the concept of integrated learning, evaluation standard, ensemble learning was divided into two stages -construction and integration of base classifiers. From the angle of bias variance decomposition analysis , It was found that the prediction accuracy of ensemble learning model could be improved by controlling the complexity of the integrated model and the difference of each base classifier. Then, we discussed the classical algorithm Bagging, Boosting and so on, and analyzed the general voting and Stacking method for the integration of classification results.. |
keywords:ensemble learning bias-variance decomposition bagging boosting Stacking |
查看全文 查看/发表评论 下载pdf阅读器 |