About this document
(Arxiv by z9rkfqhtbg is a document available to read on EtoBox.
The document presents GraphMamba, a hybrid model for hyperspectral image classification that integrates spectral-spatial token generation, graph-based token prioritization, and cross-attention mechanisms. This approach addresses challenges such as high dimensionality, spectral redundancy, and limited labeled data by effectively capturing complex spatial-spectral relationships while maintaining computational efficiency. Experimental results demonstrate that GraphMamba outperforms existing state-of-the-art mo
- Author
- z9rkfqhtbg
- Language
- EN