Deep learning has achieved remarkable success in hyperspectral image (HSI) classification, attributed to its powerful feature extraction capabilities. However, existing methods face several challenges: Convolutional Neural Networks (CNNs) are limited in modeling long-range spectral dependencies because of their limited receptive fields; Transformers are constrained by their quadratic computational complexity; and Mamba-based methods fail to fully exploit spatial–spectral interactions when handling high-dimensional HSI data. To address these limitations, we propose MRFP-Mamba, a novel Multi-Receptive-Field Parallel Mamba architecture that integrates hierarchical spatial feature extraction with efficient modeling of spatial–spectral dependencies. The proposed MRFP-Mamba introduces two key innovation modules: (1) A multi-receptive-field convolutional module employing parallel $1\times 1$, $3\times 3$, $5\times 5$, and $7\times 7$ kernels to capture fine-to-coarse spatial features, thereby improving discriminability for multi-scale objects; and (2) a parameter-optimized Vision Mamba branch that models global spatial–spectral relationships through structured state space mechanisms. Experimental results demonstrate that the proposed MRFP-Mamba consistently surpasses existing CNN-, Transformer-, and state space model (SSM)-based approaches across four widely used hyperspectral image (HSI) benchmark datasets: PaviaU, Indian Pines, Houston 2013, and WHU-Hi-LongKou. Compared with MambaHSI, our MRFP-Mamba achieves improvements in Overall Accuracy (OA) by 0.69%, 0.30%, 0.40%, and 0.97%, respectively, thereby validating its superior classification capability and robustness.

Deep metric learning (DML) has shown promising results in few-shot hyperspectral image (HSI) classification. The core idea of DML is to learn a generalized metric space in which pixels from unseen classes can be effectively classified with only a few labeled samples. However, existing DML methods mainly adopt traditional Euclidean distance to achieve the feature metric, which ignores the category uncertainty of spatial-spectral features in mixed and edge pixels. To address this issue, we fully exploit fuzzy logic theory and propose a deep fuzzy metric learning (DFML) method for few-shot HSI classification. First, we design a novel hybrid CNN-Transformer spatial-spectral feature extraction network to fully capture the spatial-spectral features of HSI pixels. Then, a fuzzy set representation method based on Gaussian membership function for spatial-spectral features is proposed, which describes the inherent fuzziness of the spatial-spectral features. Finally, to perform the fuzzy similarity measure between the fuzzy sets of query samples and prototypes, we construct a spatial-spectral fuzzy metric space, in which HSI pixels with category uncertainty in their features can be better classified under condition of small-scale labeled samples. Extensive experimental results on three public HSI datasets demonstrate that the proposed DFML method outperforms state-of-the-art few-shot HSI classification methods.

Industrial fault diagnosis is recognized as a significant technology in industrial systems. Since the breakdown events rarely happen, the scarcity of labeled fault data poses a great challenge for industrial fault diagnosis. Zero-shot learning (ZSL) has emerged as a promising approach in industrial fault diagnosis, enabling models to classify previously unseen fault data without using labeled samples. However, existing ZSL methods primarily focus on exploiting global-level fault features, but fail to capture local-level fault features that characterize the topological relationships between neighboring data points, leading to a degradation in their generalization performance. To address these issues, we propose a global-local attention-aware zero-shot learning (GLA-ZSL) method for industrial fault diagnosis. Firstly, we design a global prior refinement module (GPRM) using 2D-CNN to capture the global-level fault features. Secondly, we present a local feature enhancement module (LFEM) based on channel attention and 1D-CNN, which utilizes the extracted global-level fault features as input to deeply extract local-level fault features. Furthermore, a meta-learning strategy based on prototype distance metric for ZSL is proposed to learn more discriminative global-local fault features and improve the generalization performance of model. Finally, the proposed GLA-ZSL is evaluated on the Tennessee-Eastman process (TEP) dataset, the Case Western Reserve University (CWRU) dataset, and the Jiangnan University (JNU) bearing dataset. The experimental results demonstrate that the proposed GLA-ZSL method significantly outperforms other representative ZSL methods.

Few-shot learning (FSL) is an effective solution for cross-domain hyperspectral image (HSI) classification, which could address the limited labeled samples of the target domain. Current FSL methods mostly utilize the 3D-CNN to transform the spatial and spectral information into a single feature to model an HSI, which means that spatial and spectral information are treated equally in the feature-modeling process. However, spectral information is considered to be more domain-invariant than spatial information. Treating the spatial and spectral information equally may result in parameter redundancy and undesirable cross-domain classification performance. In this paper, we attempt to use tensor mathematics for modeling the HSI and propose a novel few-shot learning method, called tensor-based few-shot Learning (TFSL) for cross-domain HSI classification, which aims to guide the model to focus on the extraction of domain-invariant spectral dependencies. Specifically, we first propose a spatial–spectral tensor decomposition (SSTD) model to provide a mathematical explanation of the input HSI, representing the local spatial–spectral information as 1D and 2D local tensors to reduce the data redundancy. Additionally, a tensor-based hybrid two-stream (THT) model is proposed for extracting the domain-invariant spatial–spectral tensor feature by using 1D-CNN and 2D-CNN. Furthermore, we adopt a 1D-CNN tensor feature enhancement block to enhance the spectral feature of hybrid two-stream tensors and guide the THT model to concentrate on the modeling of spectral dependencies. Finally, the proposed TFSL is evaluated on four public HSI datasets, and the extensive experimental results demonstrate that the proposed TFSL significantly outperforms other advanced FSL methods.

Deep learning has demonstrated significant success in hyperspectral image (HSI) classification by effectively leveraging spatial–spectral feature learning. However, current approaches encounter three challenges: (1) high spectral redundancy and the presence of noisy bands, which impair the extraction of discriminative features; (2) limited spatial receptive fields inherent in convolutional operations; and (3) unidirectional context modeling that inadequately captures bidirectional dependencies in non-causal HSI data. To address these challenges, this paper proposes HG-Mamba, a novel hybrid geometry-aware bidirectional Mamba network for HSI classification. The proposed HG-Mamba synergistically integrates convolutional operations, geometry-aware filtering, and bidirectional state-space models (SSMs) to achieve robust spectral–spatial representation learning. The proposed framework comprises two stages. The first stage, termed spectral compression and discrimination enhancement, employs multi-scale spectral convolutions alongside a spectral bidirectional Mamba (SeBM) module to suppress redundant bands while modeling long-range spectral dependencies. The second stage, designated spatial structure perception and context modeling, incorporates a Gaussian Distance Decay (GDD) mechanism to adaptively reweight spatial neighbors based on geometric distances, coupled with a spatial bidirectional Mamba (SaBM) module for comprehensive global context modeling. The GDD mechanism facilitates boundary-aware feature extraction by prioritizing spatially proximate pixels, while the bidirectional SSMs mitigate unidirectional bias through parallel forward–backward state transitions. Extensiveexperiments on the Indian Pines, Houston2013, and WHU-Hi-LongKou datasets demonstrate the superior performance of HG-Mamba, achieving overall accuracies of 94.91%, 98.41%, and 98.67%, respectively.