Dimensional Spatial Upgrading with Variance Attention Mechanism and Its Application in the Field of Interleaved Signal Recognition

This paper proposes a new attention mechanism DSSE (Dimensional Spatial Upgrading Transformation Squeeze and Extraction Blocks) module to solve the problem of interleaved signal recognition. The DS-SE module addresses the challenge of how to introduce attention into one-dimensional data sets by means of dimensional spatial transformation. The experimental results show that the DS-SE module performs better than the traditional residual convolutional neural network and SENeT in a series of interleaved signal data sets. The introduction of the DSSE module enables the network to better extract fine-grained features, thereby enhancing its representational capacity.