一种基于降噪自编码的组合分类算法

Ensemble Learning Based on Denosing Autoencoder

针对传统分类学习算法的准确性现状进行了研究,提出了一种基于降噪自编码的组合分类算法(Ensemble Learning based on Denosing Autoencoder,ELDA).与Bagging、Adaboost以及Rotation Forest等传统的组合分类器学习方法不同,ELDA首先通过使用降噪自编码算法将数据集映射到新的特征空间,然后在此空间学习得到决策树作为基分类器,最后对数据集进行类别预测.通过与Bagging、Adaboost及Rotation Forest学习方法相比,结果表明:ELDA在预测精度上显著优于对比算法.

A ensemble learning method(Ensemble Learning based on Denosing Autoencoder,ELDA)is proposed to study the accuracy of traditional classification learning algorithm.Unlike traditional ensemble learning approaches such as Bagging,Adaboost and Rotation Forest,ELDA first maps the data sets to a new feature space by using denoising autoencoder algorithm,then constructs the decision tree as the base classifier,and further classifies the data sets.Experimental results show that the accuracy of ELDA is higher than others,and it is proved that ELDA is an effective classifier ensemble algorithm of denosing autoencoder.