Exploitation of Ephemeral Features in Sonar Classification Algorithms

For active and passive sonar, there is a rich history of classification strategies and associated derived contact features.Ā However, a demonstrated need persists for robust classification in challenging operating environments where targets may only exist for fleeting moments or where targets may intentionally take deceptive and evasive measures that limit detection.Ā During the Phase II effort, Etegent Technologies Ltd. will continue to develop and mature an ephemeral feature technology that provides a robust classification system for missing and ephemeral contact features.Ā The proposed framework leverages a combination of Bayesian and machine learning methods to optimally exploit physical and virtual information sources while maintaining robustness to limited contact information from fleeting and deceptive targets.