基于支持向量机的水声信号多分类器设计

目标分类器是水下目标自动识别系统的重要组成部分，目前水下目标分类的方法主要有统计分类、神经网络和专家系统等三大类的分类方法。支持向量机（SVM，Support Vector Machine）是根据统计理论提出的一种新的算法，该算法具有良好的泛化性能，不仅对训练样本的分类性能较好，对未知的检验样本同样具有好的分类效果，特别适用于小样本数据的分类。本文将该算法推广至多分类情况，并对三类水声信号样本进行分类试验。实验结果表明，该算法可以有效的避免“维数灾难”问题，且分类正确率高于传统的神经网络分类器。

Multiple classifier design of underwater acoustic signal based on SVM

Target classification is an important part of the underwater target automatic identification system, the current method of classification of underwater targets are mainly three categories of classification statistical classification, neural networks and expert systems. Support vector machine （SVM, Support Vector Machine） is a new algorithm based on statistical theory, that the algorithm has good generalization performance, not only for better classification performance of training samples, testing unknown samples also has good classification results, especially for small samples of data classification. In this paper, the algorithm is extended up to multiple classification, and testing three types of underwater acoustic signal. Experimental results show that the algorithm can effectively avoid the＆quot;dimensional disaster＆quot;and the correct classification rate is higher than the traditional neural network classifier.