Unlocking the Potential of Data Augmentation in Contrastive Learning for Hyperspectral Image Classification

The lack of training samples remains one of the major obstacles in applying convolutional neural networks (CNNs) to the hyperspectral image (HSI) classification. In this letter, the accurate classification of HSI with limited training samples is investigated. Due to the advantages of minimizing the distance between samples in the same class and maximizing the distance between samples in different classes, siamese CNN is used for HSI classification with limited training samples. After that, to improve the classification performance, data augmentation is investigated for siamese CNN-based HSI classification. Specifically, pair data augmentation based on CutMix is proposed to generate the training pairs of the same or different classes and the new generated training pairs are used to train siamese CNN. Traditional data augmentation methods simply generate new training samples. It is not proper when data augmentation methods keep the loss function and change the content of the input at the same time. Therefore, a soft-loss-based siamese CNN, which changes its loss according to the coupled replacement data augmentation, is proposed to further address the HSI classification with limited training samples. In the experimental part, on two widely used hyperspectral datasets, the influences of different training samples and window sizes are discussed, and the experimental results reveal that the proposed soft-augmentation-based siamese CNN provides competitive results with limited training samples compared with state-of-the-art methods.

Hyperspectral image (HSI) classification has become a hot topic in the field of remote sensing. In general, the complex characteristics of hyperspectral data make the accurate classification of such data challenging for traditional machine learning methods. In addition, hyperspectral imaging often deals with an inherently nonlinear relation between the captured spectral information and the corresponding materials. In recent years, deep learning has been recognized as a powerful feature-extraction tool to effectively address nonlinear problems and widely used in a number of image processing tasks. Motivated by those successful applications, deep learning has also been introduced to classify HSIs and demonstrated good performance. This survey paper presents a systematic review of deep learning-based HSI classification literatures and compares several strategies for this topic. Specifically, we first summarize the main challenges of HSI classification which cannot be effectively overcome by traditional machine learning methods, and also introduce the advantages of deep learning to handle these problems. Then, we build a framework that divides the corresponding works into spectral-feature networks, spatial-feature networks, and spectral-spatial-feature networks to systematically review the recent achievements in deep learning-based HSI classification. In addition, considering the fact that available training samples in the remote sensing field are usually very limited and training deep networks require a large number of samples, we include some strategies to improve classification performance, which can provide some guidelines for future studies on this topic. Finally, several representative deep learning-based classification methods are conducted on real HSIs in our experiments.

Few-shot learning (FSL) uses prior knowledge and supervised experience to effectively classify hyperspectral images (HSIs), thereby reducing the cost of large numbers of labeled samples. However, existing few-shot methods ignore the correlation between cross-domain feature channels, and the feature representation ability is insufficient. To address above issue, this paper proposes a novel Residual Channel Attention Based Sample Adaptation Few-Shot Learning for Hyperspectral Image Classification(RCASA-FSL) for hyperspectral image classification (HSIC), which can capture and enhance cross-domain dependencies through multi-layer residual connection and random-based feature recalibration. Specifically, a Deep Residual Feature Channel Attention Mechanism (DRFCAM) is designed to obtain cross-domain dependencies by residual concatenation, and further the residual structure is stacked for mining depth discrimination information. Furthermore, a new Random-based Feature Recalibration Module (RFRM) is proposed to reassign the feature weights via random matrix, which fully explore feature weight relationships to guide the sample adaptation process. Besides, we design a joint loss function with combining the FSL loss and domain adaptive loss for further optimization model. Experiments conducted on several standard hyperspectral datasets demonstrate that the proposed RCASA-FSL is superior to other FSL techniques in both quantitative and qualitative aspects.

The vision transformer (ViT) has become a hot topic in image processing due to its global feature extraction capabilities. However, the ViT suffers from over-smoothing in feature extraction and over-fitting in the training procedure, so it is hard to achieve satisfactory performance in hyperspectral image (HSI) classification. To address these issues, we propose a ViT with contrastive learning (CViT). The network architecture includes a patch embedding module, transformer blocks, and a classifier. The training of CViT can be considered as an optimization problem with a supervised contrastive loss, an unsupervised contrastive loss, and an ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> -regularizer with respect to linear self-attention weights. Specifically, the supervised contrastive loss is proposed to alleviate the negative effects of HSI features’ spectral variability and spatial diversity by increasing intra-class consistency. On the other hand, the unsupervised contrastive loss is exploited to reduce redundancy by reconstructing global structural information. In particular, regularized linear self-attention weights reduce the over-smoothing issue. Extensive experimental results on three HSI datasets demonstrate that the proposed CViT achieves competitive performance.

Class incremental learning with analytic learning for hyperspectral image classification

Intelligent fault diagnosis is an important method in rotating machinery fault diagnosis and equipment health management. To deal with co-frequency vibration faults, a type of typical fault in rotating machinery, this paper proposes a fault diagnosis method based on the stacked autoencoder (SAE) and ensembled ResNet-SVM. Furthermore, the time- and frequency-domain features of several co-frequency vibration faults are summarized based on the mechanism analysis and calculated using actual vibration data. To realize and validate the high-precision diagnosis method of rotating equipment with co-frequency faults proposed in this study, the following three criteria are required: First, to improve the effectiveness and robustness of the ensembled model and the sliding window using data augmentation, adding noise, autoencoder (AE) and SAE methods are analyzed in terms of principle and practical effects. Second, ResNet is used as the feature extractor for the ensembled ResNet-SVM model. Feature extraction is carried out twice, and the extracted co-frequency fault features are more comprehensive. Finally, the data augmentation method and ensemble ResNet-SVM are combined for fault diagnosis and compared with other methods. The experimental results show that the accuracy of the proposed method can exceed 99.9%.

Hyperspectral image (HSI) classification remains a challenging problem due to unique characteristics of HSI data (such as numerous bands and strong correlations in the spectral and spatial domains) and small sample size. To address such concerns, we propose a novel spectral–spatial feature extraction method for HSI classification by employing graph embedding and deep learning (DL) models. Since the conventional graph cannot capture the complex manifold relationship of HSI data, and there exist the observations of within-class variation as well as the similarity between different classes in the spectral domain, we construct the supervised within-class/between-class hypergraph (SWBH) to extract the spectral features of HSI. Since it is difficult for DL models to learn representative features for HSI data when the labeled training samples are limited, we propose the random zero settings to newly generate a large amount of labeled HSI samples for the training of convolutional neural network (CNN). The designed sample expanded CNN (SECNN) is used to extract the HSI spatial features. Thus, the spectral–spatial features of HSI can be learned by integrating the features extracted from SWBH and SECNN, respectively. Experiments on three real HSI datasets demonstrate higher classification accuracy of the proposed SWBH–SECNN method.

Hyperspectral image (HSI) classification accuracy has been greatly improved by employing deep learning. The current research mainly focuses on how to build a deep network to improve the accuracy. However, these networks tend to be more complex and have more parameters, which makes the model difficult to train and easy to overfit. Therefore, we present a lightweight deep convolutional neural network (CNN) model called S2FEF-CNN. In this model, three S2FEF blocks are used for the joint spectral–spatial features extraction. Each S2FEF block uses 1D spectral convolution to extract spectral features and 2D spatial convolution to extract spatial features, respectively, and then fuses spectral and spatial features by multiplication. Instead of using the full connected layer, two pooling layers follow three blocks for dimension reduction, which further reduces the training parameters. We compared our method with some state-of-the-art HSI classification methods based on deep network on three commonly used hyperspectral datasets. The results show that our network can achieve a comparable classification accuracy with significantly reduced parameters compared to the above deep networks, which reflects its potential advantages in HSI classification.

In Hyperspectral Image (HSI) classification, acquiring large quantities of high-quality labeled samples is typically costly and impractical. Traditional deep learning methods are limited in such scenarios due to their dependence on sample quantities. To address this challenge, researchers have turned to Few-Shot Learning (FSL). Although existing FSL methods improve classification performance by enhancing domain invariance through domain adaptation, they often overlook the critical issue of high inter-class similarity and large intra-class variability. Moreover, during domain alignment, features of different categories within the same domain may become confused. To address these issues, this paper proposes a novel Domain-Invariant Few-Shot Contrastive Learning (DIFSCL) method, which combines domain adaptation and contrastive learning strategies to not only learn domain-invariant features but also significantly enhance inter-class discriminability. Based on this, we further design a multi-scale adaptive attention mechanism for a hyperspectral feature extraction network to more effectively extract and optimize generalized features within the DIFSCL framework, significantly improving intra-class consistency and inter-class discriminative capability of features. Experimental results on three widely used HSI datasets demonstrate that our method significantly outperforms existing techniques in few-shot classification tasks.

Hyperspectral image classification is an important topic in the field of remote sensing. However, the high dimensionality and high spatial-spectral correlation of hyperspectral image will easily lead to poor classification performance due to the Hughes phenomenon. In this paper, we proposed a adaptive group local Riemannian embedding, called AGLRE, to extract the spatial and spectral features from hyperspectral image. It firstly mapped original data into a Riemannian manifold by constructing region covariance matrices for each pixel of hyperspectral image. And the multiple local tangent space on Riemannian manifold were learned by adaptive neighbourhood strategy. Lastly, the local linear embedding was applied to reduce the dimensionality and merge multiple tangent space into a global coordinate. Experimental results on public hyperspectral data set showed that the proposed method can achieve higher classification performance than other competing algorithms.

The rapid serial visual presentation-based brain–computer interface (RSVP-BCI) system achieves the recognition of target images by extracting event-related potential (ERP) features from electroencephalogram (EEG) signals and then building target classification models. Currently, how to reduce the training and calibration time for classification models across different subjects is a crucial issue in the practical application of RSVP. To address this issue, a zero-calibration (ZC) method termed Attention-ProNet, which involves meta-learning with a prototype network integrating multiple attention mechanisms, was proposed in this study. In particular, multiscale attention mechanisms were used for efficient EEG feature extraction. Furthermore, a hybrid attention mechanism was introduced to enhance model generalization, and attempts were made to incorporate suitable data augmentation and channel selection methods to develop an innovative and high-performance ZC RSVP-BCI decoding model algorithm. The experimental results demonstrated that our method achieved a balance accuracy (BA) of 86.33% in the decoding task for new subjects. Moreover, appropriate channel selection and data augmentation methods further enhanced the performance of the network by affording an additional 2.3% increase in BA. The model generated by the meta-learning prototype network Attention-ProNet, which incorporates multiple attention mechanisms, allows for the efficient and accurate decoding of new subjects without the need for recalibration or retraining.

Variations in atmosphere, lighting, and imaging systems result in diverse category distributions in hyperspectral imagery, impacting the accuracy of cross-domain hyperspectral image classification (HSIC). Unsupervised domain adaptation (UDA) aims to address this issue by learning a model that generalizes effectively across domains, leveraging labels only from source domain (SD). Most existing UDA methods focus on aligning distributions between domains without fully considering the valuable information within individual domains. To fill this gap, this letter proposes a domain-collaborative contrastive learning (DCCL) method. DCCL integrates a novel pseudo-labeling strategy with a cross-domain contrastive learning (CL) framework. Specifically, in the pseudo-labeling phase, the confident examples in target domain (TD) are collaboratively labeled according to the labeled examples in SD and the class centers in TD. Then, the CL phase simultaneously minimizes in-domain and cross-domain contrastive loss to promote the aggregation of examples from the same category in both domains. Experimental results demonstrate that the DCCL achieves the accuracy rates of 93.47% and 54.59% on Pavia and Indiana datasets, respectively, surpassing the performance of other state-of-the-art UDA methods. Our source code is available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/Leap-luohaiyang/DCCL-2024</uri>.

Multiclass feature learning for hyperspectral image classification: Sparse and hierarchical solutions

In this work, we propose a new semisupervised active learning approach for hyperspectral image classification. The proposed method aims at improving machine generalization by using pseudolabeled samples, both confident and informative, which are automatically and actively selected, via semisupervised learning. The learning is performed under two assumptions: a local one for the labeling via a superpixel-based constraint dedicated to the spatial homogeneity and adaptivity into the pseudolabels, and a global one modeling the data density by a multinomial logistic regressor with a Markov random field regularizer. Furthermore, we propose a density-peak-based augmentation strategy for pseudolabels, due to the fact that the samples without manual labels in their superpixel neighborhoods are out of reach for the automatic sampling. Three real hyperspectral datasets were used in our experiments to evaluate the effectiveness of the proposed superpixel-based semisupervised learning approach. The obtained results indicate that the proposed approach can greatly improve the potential for semisupervised learning in hyperspectral image classification.

Deep learning has recently been applied to automatically classify the modulation categories of received radio signals without manual experience. However, training deep learning models requires massive volume of data. An insufficient training data will cause serious overfitting problem and degrade the classification accuracy. To cope with small dataset, data augmentation has been widely used in image processing to expand the dataset and improve the robustness of deep learning models. However, in wireless communication areas, the effect of different data augmentation methods on radio modulation classification has not been studied yet. In this paper, we evaluate different data augmentation methods via a state-of-the-art deep learning-based modulation classifier. Based on the characteristics of modulated signals, three augmentation methods are considered, i.e., rotation, flip, and Gaussian noise, which can be applied in both training phase and inference phase of the deep learning algorithm. Numerical results show that all three augmentation methods can improve the classification accuracy. Among which, the rotation augmentation method outperforms the flip method, both of which achieve higher classification accuracy than the Gaussian noise method. Given only 12.5% of training dataset, a joint rotation and flip augmentation policy can achieve even higher classification accuracy than the baseline with initial 100% training dataset without augmentation. Furthermore, with data augmentation, radio modulation categories can be successfully classified using shorter radio samples, leading to a simplified deep learning model and shorter the classification response time.