The underwater acoustic recognition network based on dynamic attention

This paper proposes an innovative underwater acoustic target recognition network framework, BiTransKan, which integrates Dynamic Sparse Attention (BiFormer), a Dual Dynamic Mixing Module (D-Mixer), and interpretable Kolmogorov-Arnold Networks (KAN), significantly improving recognition efficiency and accuracy in complex underwater acoustic environments. Experimental results based on the DeepShip dataset show that the proposed model outperforms existing methods in terms of accuracy, computational efficiency, and model interpretability, providing a new technical solution and development direction for future research and practical applications in underwater acoustic target recognition.