如何结合两个机器学习算法的输出? [英] How to combine two machine learning algorithm outputs?

问题描述

如果我有两个截然不同的数据集和两种截然不同的分类技术，是否有很好的方法来组合两个输出?我知道平均水平可能有效，但是还有其他更相关的方法吗?我听说过一些概念，例如促进学习和整体学习，这些概念是否适用?

If I have two very different datasets and two very different classification techniques, is there a good way to combine the two outputs? I understand an average may work but is there a more relevant way to do this? I've heard of several concepts like boosting and ensemble learning, would these be applicable?

推荐答案

有两种一般方法可以解决此问题.第一个称为 boosting ，它使用加权投票来决定预测.主要思想是结合两个分类器的优势.

There are two general ways to go about this problem. The first, called boosting, uses weighted voting to decide on the prediction. The main idea is to combine advantages of both classifiers.

第二种方法称为 stacking ，它使用两个分类器的输出作为特征进入另一个分类器(可能具有其他特征，例如原始特征)，并将最终分类器的输出用于预测.

The second approach, called stacking, uses the outputs of the two classifiers as features into another classifier (possibly with other features, e.g. the original ones), and use the output of the final classifier for the prediction.

在没有更多细节的情况下，这是我能给出的最佳答案.

In the absence of further details, this is the best answer I can give.

请参见机器学习中的打包，增强和堆叠 Stats.SE上的更多内容.

See Bagging, boosting and stacking in machine learning on Stats.SE for more.

这篇关于如何结合两个机器学习算法的输出?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！