深层次特征学习的Adaboost大规模图像分类算法

Adaboost image classification based on deep feature learning

针对浅层次大规模图像分类的低精度问题,提出深层次特征学习的Adaboost图像分类算法。首先以DBN作为弱分类器对样本图像进行学习,根据每次训练得到的分类错误率以及各样本的分类准确性调整权值;然后在所有弱分类器训练好以后,使用BP算子回溯再次整体调整体样本权值;最后将所有弱分类器集成强分类器,输出最终分类结果。使用MNIST和ETH-80两种数据集进行实验仿真,并将分类结果与其他算法进行比较。结果表明所提算法的分类精度明显高于其他算法,有效实现了高精度的大规模图像分类。

Aiming at the low precision problem of shallow large scale image classification,this paper proposes an adaboost algorithm in large-scale image classification based on deep feature learning.Firstly,DBN is used as the weak classifier to learn the sample images,and the weights are adjusted according to the classification error rate and the classification accuracy of each sample.After all the weak classifiers are trained,the BP operator is used to readjust the sample weight and output the final error rate of each classifier.Finally,all weak classifiers are integrated into strong classifier and output the final results.This paper simulates the experiment with two kinds of data sets,which are MNIST and ETH-80.Comparing the classification results to other algorithms,the classification accuracy of this algorithm is higher than others.High precision image classification is realized.