《计算机应用研究》|Application Research of Computers

基于深度玻尔兹曼机的乐器分类问题研究

Research on musical instrument classification based on deep Boltzmann machine

免费全文下载 (已被下载 次)  
获取PDF全文
作者 周畅,米红娟
机构 兰州财经大学 信息工程学院,兰州 730020
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2019)07-026-2031-04
DOI 10.19734/j.issn.1001-3695.2018.01.0022
摘要 应用传统浅层模型处理乐器分类任务存在非线性拟合能力较差的问题,使分类准确率得不到有效保证,有必要引入深度学习方法来提升复杂任务的非线性建模能力。将深度玻尔兹曼机作为特征提取器提取表达能力更强的数据特征,分别以SVM与softmax分类器作为深度神经网络的顶层设置形成DBM+SVM组合模型与DBM+softmax组合模型,引入平均场理论和动量项因子优化网络训练过程。将上述两组模型及单一SVM分类器在五类乐器音频数据上进行对比实验,两种深度学习组合模型的分类准确率分别达到89.29%和87.5%,与传统浅层分类方法SVM的73.21%的准确率相比优势明显。实验结果表明深度玻尔兹曼机在乐器分类领域的应用颇具前景。
关键词 深度玻尔兹曼机; 乐器分类; 深度学习; 平均场理论; 动量项
基金项目
本文URL http://www.arocmag.com/article/01-2019-07-026.html
英文标题 Research on musical instrument classification based on deep Boltzmann machine
作者英文名 Zhou Chang, Mi Hongjuan
机构英文名 College of Information Engineering,Lanzhou University of Finance & Economics,Lanzhou 730020,China
英文摘要 The application of traditional shallow model to instrument classification task has the problem of poor nonlinear fitting ability, so that the accuracy of classification was not guaranteed effectively. It is necessary to introduce deep learning method to improve the nonlinear modeling ability of complex tasks. This paper used deep Boltzmann machine as feature extractor to abstract more expressive deep learning features. It respectively used SVM and softmax classifier as top layer of deep neural network to form DBM+SVM and DBM+softmax combined model. Besides, this paper introduced the mean field theory and momentum factor to optimize the network training process, and compared the above two sets of models and single SVM classifier on 5 kinds of musical instruments audio data. The classification accuracy of the two types of deep learning combination models reached 89.29% and 87.5% respectively, compared with the accuracy of the traditional shallow classification method SVM of 73.21%. The experimental results show that the application of deep Boltzmann machine in the field of musical instrument classification is very promising.
英文关键词 deep Boltzmann machine(DBM); instrument classification; deep learning; mean field theory; momentum term
参考文献 查看稿件参考文献
 
收稿日期 2018/1/13
修回日期 2018/3/20
页码 2031-2034
中图分类号 TP389.1
文献标志码 A