This study provides a unique opportunity for advancements in underwater acoustic target recognition (UATR). The ShipsEar dataset, comprised of underwater acoustic recordings captured by hydrophones across eight distinct locations, encapsulates diverse environmental conditions and vessel sound characteristics. This study leverages meta-learning techniques for domain adaptation to enhance the generalizability of machine-learning models in cross-location vessel recognition tasks. By analyzing inter-location variations and identifying domain-invariant features, the proposed approach addresses key challenges in underwater acoustic signal processing, such as variability in ambient noise and recording conditions. The findings have potential applications in maritime safety, environmental monitoring, and naval operations. This work was supported by the Korea Research Institute for defense Technology planning and advancement (KRIT) grant funded by the Korean government (DAPA—Defense Acquisition Program Administration) (No. 21-107-B00-008(KRIT-CT-23-009), The technical research for Underwater surveillance in open-sea and the analytic technique of acoustical environments in deep water, 2025)
Bae et al. (Tue,) studied this question.