Feature Extraction of Active Sonar Data using Independent Component Analysis

摘要

Independent component analysis (ICA) is a statistical signal processing method for estimating the source signals from the mixed observation data. In this paper, however, we apply the ICA model to the training data set and use each row vector of the mixing matrix in the ICA model as a new feature vector for sonar data classification. The sonar data set that has been released from UCI machine learning repository is used for classification experiments. The support vector machine is used as a binary classifier. Experimental results are presented with our discussions.