MambaMoE: Mixture-of-Spectral-Spatial-Experts State Space Model for Hyperspectral Image Classification

2025-04-29

AI Summary

Existing Mamba-based models for hyperspectral image (HSI) classification overlook spectral-spatial directional heterogeneity, limiting discriminative capability in complex heterogeneous regions. Method: This paper proposes the first hyperspectral-specific Mixture-of-Experts (MoE) spectral-spatial adaptive modeling framework. It introduces (1) a sparsely activated Spectral-Spatial Mixture-of-Experts Block (MoMEB) to explicitly model spectral and spatial features with directional specialization; (2) an Uncertainty-Guided Correction Learning (UGCL) strategy to enhance robustness against heterogeneous regions; and (3) a hybrid architecture integrating state-space modeling with MoE to jointly capture long-range dependencies and maintain computational efficiency. Results: Evaluated on multiple benchmark HSI datasets, the method achieves state-of-the-art classification accuracy and inference efficiency with significantly fewer parameters than mainstream Mamba- and Transformer-based baselines, demonstrating superior generalization and scalability.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Computer Vision: Multi-modal Vision
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

The Mamba model has recently demonstrated strong potential in hyperspectral image (HSI) classification, owing to its ability to perform context modeling with linear computational complexity. However, existing Mamba-based methods usually neglect the spectral and spatial directional characteristics related to heterogeneous objects in hyperspectral scenes, leading to limited classification performance. To address these issues, we propose MambaMoE, a novel spectral-spatial mixture-of-experts framework, representing the first MoE-based approach in the HSI classification community. Specifically, we design a Mixture of Mamba Expert Block (MoMEB) that leverages sparse expert activation to enable adaptive spectral-spatial modeling. Furthermore, we introduce an uncertainty-guided corrective learning (UGCL) strategy to encourage the model's attention toward complex regions prone to prediction ambiguity. Extensive experiments on multiple public HSI benchmarks demonstrate that MambaMoE achieves state-of-the-art performance in both accuracy and efficiency compared to existing advanced approaches, especially for Mamba-based methods. Code will be released.

Problem

• Enhancing HSI classification by addressing spectral-spatial directional neglect
• Introducing first MoE-based framework for adaptive spectral-spatial modeling
• Improving ambiguous region prediction via uncertainty-guided corrective learning

Innovation

• Spectral-spatial mixture-of-experts framework for HSI
• Mixture of Mamba Expert Block for adaptive modeling
• Uncertainty-guided corrective learning for complex regions