Accurate deep and direction classification model based on the antiprism graph pattern feature generator using underwater acoustic for defense system

Abstract

Underwater acoustic is one of the hot-topic and complex research areas for advanced signal processing. In this research, our main motivation is to recommend a high accurate underwater sound classification method using a special graph-based feature generator. The most valuable features have been selected using ReliefF iterative neighborhood component analysis (RFINCA) selector. In the classification phase, Decision Tree (DT), k nearest neighbor (kNN), Linear Discriminant (LD), Naive Bayes (NB), and support vector machine (SVM) classifiers have been used with 10-fold cross-validation. To calculate the performance of the TQWT and antiprism graph pattern-based feature generation and RFINCA selector-based sound classification method, two underwater acoustic datasets have been collected. According to tests, the best accurate classifier is SVM. SVM attained 90.33% and 96.91% accuracies for the collected depth and direction datasets respectively. The calculated results denoted the success of the presented antiprism graph pattern-based method for underwater acoustic classification.