Neural Networks for Active Sonar Classification

me active sonar classijication has been a challenging pattern recognition problem for many years mainly due to the complexity of ocean environment. Improvement of sensors and data acquisition can be very costly and can only provide limited improvement in classgcation. Traditional classijiers are not easily adaptable to the changing ocean characteristics. Neural networks which experienced a major progress in recent years are ideally suited for the active sonar classijication problems with the potential advantages: highly adaptive to the changing ocean environment, possibly very signijicant improvement in recognition rate and speed, better performance at low signal-to-noise ratio, and less sensitive to small training size, etc. Neural networks indeed can be considered as the major development in pattern recognition in recent years. In this paper, some active sonar data characteristics are presented, and the peformances of several feedforward neural networks are evaluated and compared with the traditional nearest neighbor decision rule. It is concluded that the neural networks studied not only can outper$orm but also are far more robust than the traditional classifiers.