论文部分内容阅读
针对多媒体信息中的音频信号,提出一种基于线性判别分析(LDA)与极限学习机(ELM)的分类方法.首先,使用傅里叶变换等方法从每一段音频中提取特征,并将它们按比例组成一个高维向量;其次,应用LDA对高维向量进行降维,使其成为用于分类的最优特征,作为ELM的训练和测试样本;最后,分别采用ELM,SVM,BP分类器对4种音频信号进行分类,并进行性能对比与分析.实验表明:提出的算法对于较难分的类也具有较好的分类效果,平均正确率为90%,同时运算速度比SVM快一千多倍.
Aiming at the audio signal in multimedia information, a classification method based on Linear Discriminant Analysis (LDA) and Extreme Learning Machine (ELM) is proposed.Firstly, features such as Fourier transform are used to extract features from each segment of audio and press Secondly, the LDA is used to reduce the dimensionality of the high-dimensional vector, making it the best feature for classification as training and testing samples of ELM. Finally, ELM, SVM and BP classifier are used respectively 4 kinds of audio signals are classified and their performances are compared and analyzed.Experiments show that the proposed algorithm has better classification results for the harder classes with the average correct rate of 90% and the speed of operation more than a thousand times faster than SVM Times