Knowledge distillation essentially maximizes the mutual information between teacher and student networks. Typically, a variational distribution is introduced to maximize the variational lower bound. However, the heteroscedastic noises derived from this distribution are often unstable, leading to unreliable data-uncertainty modeling. Our research identifies that bias-variance coupling in knowledge distillation causes this instability. We thus propose Bias-variance dEcomposition kNowledge dIstillatioN (BENIN) approach. Initially, we use bias-variance decomposition to decouple these components. Subsequently, we design a lightweight Feature Frequency Expectation Estimation Module (FF-EEM) to estimate the student's prediction expectation, which helps compute bias and variance. Variance learning measures data uncertainty in the teacher's prediction. A balance factor addresses the bias-variance dilemma. Lastly, the bias-variance decomposition distillation loss enables the student to learn valuable knowledge while reducing noise. Experiments on Synapse and Lits17 medical-image-segmentation datasets validate BENIN's effectiveness. FF-EEM also mitigates high-frequency noise from high mask rates, enhancing data-uncertainty estimation and visualization. Our code is available at https://github.com/duanzhongjian/BENIN .

Representing hyperspectral images (HSI) is a complex and challenging task, primarily due to spectral uncertainty. Learnable prototypical contrastive learning is specialized in discriminative instance representation. However, it requires a much low temperature to prevent model collapse, which can hinder the encoder’s ability to capture category relationships. Furthermore, self-supervision at network terminal could obscure specific semantics in hyperfine spectra. In this paper, we propose a hierarchical and bidirectional learnable prototypical contrastive learning method (HiBiCo) for HSI representation. We build learnable prototype dictionaries at shallow and final layers of the network for deep contrastive supervision. Moreover, we introduce reverse contrastive learning under negative sample dominance to address excessively uniform representations caused by traditional positive-dominated contrastive loss in dual-dictionary hierarchical supervision. By allowing deep contrastive supervision and two-way information flow along with the general InfoNCE loss, our approach alleviates the uniform distribution, maintains the latent data manifold, and enables diverse and effective representation. Experiments with linear probing demonstrate the effectiveness of our HiBiCo framework in handling complex scenes, highlighting the potential of self-supervised pretraining for hyperspectral image representation. The code is available at https://github.com/sakurashine/HiBiCo.

Hyperspectral images provide plentiful latent information that requires exploration for ground object recognition, where self-supervised learning is efficient and independent of manual labeling. However, the severe spectral uncertainty poses a significant challenge in discriminative and generalizable representation by self-supervision. This letter proposes a variational generative transformer with momentum contrastive supervision (ConVaT) to alleviate the problem. ConVaT contains two branches: a variational generative branch and a contrastive learning branch—the former guides informative data representation via an encoder-decoder transformer with variational inference; the latter encourages the representation with discriminability by distinguishing positive anchors from negative ones. Significantly, to facilitate a more generalizable latent representation, we reconstruct data with reparameterized tokens sampled multiple times from the global anchor, instead of the latent representation of unmasking data. Extensive experiments on three public datasets show that ConVaT is superior in data representation with intra-class clustering and inter-class distinction, and it achieves considerable improvements over present methods under linear probing, especially for the Indian pines dataset with intense spectral uncertainty. Our code will be available at https://github.com/liuzuo-byte/ConVaT.

The global modeling capability of vision transformer (ViT) has been well proven in the field of hyperspectral image (HSI) classification. However, ViT does not have the excellent local feature extraction capability compared with the convolutional neural network (CNN). Therefore, early-stage convolutions are often used to enhance ViT’s local representation ability. However, directly applying convolutions on high-dimensional HSI data increases computational overhead. Moreover, recent researches have observed that ViT may suffer from high redundancy in capturing multihead self-attention (MHSA). To address the above issues, we propose a multiscale super token transformer (MSSTT) model for HSI classification. We use a divide-and-conquer strategy to extract local features and global dependencies of HSI data at multiple granularities. Specifically, our proposed model incorporates two branches: a multiscale convolution (MSConv) branch that uses various convolutional kernels to extract diverse local features and a multiscale super token attention (MSSTA) branch for capturing global features with low redundancy. Finally, comparative experimental results with advanced methods show that the proposed MSSTT possesses better classification performance. On the Salinas (SA), Pavia University (PU), and Kennedy Space Center (KSC) datasets, the overall accuracies (OAs) of our MSSTT are 98.47%, 98.47%, and 99.38%, respectively. Code will be released athttps://github.com/zhangtaizheng/MSSTT.

Three-dimensional target reconstruction from 2-D synthetic aperture radar (SAR) images can reduce the difficulty of data acquisition when compared to 3-D imaging technologies. Pixel2mesh was well designed for 3-D target reconstruction from 2-D optical images. However, when it is directly applied to small SAR datasets, overfitting is prone to occur. In this letter, we propose a lightweight Pixel2mesh for 3-D target reconstruction from a single 2-D SAR image. Based on the original Pixel2mesh, we improve two subnetworks: feature extraction and 3-D deformation. First, we lightweight the graph residual network (G-ResNet) module in the 3-D deformation subnetwork to avoid overfitting. Second, we add a decoder after the feature extraction subnetwork to reconstruct the input image and then obtain the image reconstruction loss to guide the training of the whole network. In addition, we modify the Laplace regularization term for the training of the proposed network, aiming to make the deformation more reasonable. Experiments are implemented on the Gotcha dataset, where 2-D images of seven cars are obtained by performing 2-D imaging. The 3-D labels of these cars are generated by using their CAD models downloaded from public websites. Experimental results verify that our network can achieve better 3-D reconstruction results than the original Pixel2mesh.

Contrastive learning with learnable examples performs outstandingly in data representation. However, when dealing with hard samples, instance-level alignment with excessive uniformity may descend into trivial clusters, especially when confronted with inter-class similarity and intra-class diversity in hyperspectral images. To solve this problem, we regard prototypical contrastive learning as tracing the potential probability density distribution. Then, a novel pre-training method, Learnable Sparse Contrastive Sampling (LSCoSa), is proposed for discriminative representation learning, containing sparse positive sampling and multiple positives learning. Specifically, on the basis of cooperative-adversarial contrastive learning, we first exert a KL divergence regularizer on the average activation probability of the prototypes, suppressing fake density prototypes for sparse positive sampling. Furthermore, we propose multiple positives learning, in which the top-k potential positives are retrieved and dynamically weighted for contrastive supervision, to avoid trivial clusters and cover satisfying semantic variations. Comprehensive experiments on three HSI benchmark datasets demonstrate that LSCoSa achieves significant advantages over other HSIC methods. The code is available at https://github.com/sakurashine/LSCoSa.

Convolutional neural networks (CNNs) showed impressive performance for hyperspectral image (HSI) classification. Nevertheless, convolutional layers contain massive parameters, which restrict the deployment of CNNs on satellite and airborne platforms with limited storage and computing resources. In this letter, we propose a lightweight spectral-spatial convolution module (LS2CM) as an alternative to the convolutional layer. The proposed LS2CM can greatly reduce network parameters and computational complexity in terms of multiply-accumulate operations (MACs) while maintaining or even improving the classification performance. Furthermore, it is a plug-and-play component and can be used to upgrade existing CNN-based models for HSI classification. Experimental results on two benchmark HSI data sets demonstrate that the proposed LS2CM achieves competitive results in comparison with other state-of-the-art methods.

A growing number of prosumers have entered the local power market in response to an increase in the number of residential users who can afford to install distributed energy resources. The traditional microgrid trading platform has many problems, such as low transaction efficiency, the high cost of market maintenance, opaque transactions, and the difficulty of ensuring user privacy, which are not conducive to encouraging users to participate in local electricity trading. A blockchain‐based mechanism of microgrid transactions can solve these problems, but the common single‐blockchain framework cannot manage user identity. This study thus proposes a mechanism for secure microgrid transactions based on the hybrid blockchain. A hybrid framework consisting of private blockchain and consortium blockchain is first proposed to complete market transactions. The private blockchain stores the identifying information of users and a review of their transactions, while the consortium blockchain is responsible for storing transaction information. The block digest of the private blockchain is stored in the consortium blockchain to prevent information on the private blockchain from being tampered with by the central node. A reputation evaluation algorithm based on user behavior is then developed to evaluate user reputation, which affects the results of the access audit on the private blockchain. The higher a user’s reputation score is, the more benefits he/she can obtain in the transaction process. Finally, an identity‐based proxy signcryption algorithm is proposed to help the intelligent management device with limited computing power obtain signcryption information in the transaction process to protect the transaction information. A system analysis showed that the secure transaction mechanism of the microgrid based on the hybrid blockchain boasts many security features, such as privacy, transparency, and imtamperability. The proposed reputation evaluation algorithm can objectively reflect all users’ behaviors through their reputation scores, and the identity‐based proxy signcryption algorithm is practical.

Deep multiscale features extracted from diverse perspectives present a more powerful ability than shallow ones for hyperspectral image (HIS) classification. In this letter, we proposed a deep feature-based multitask joint sparse representation (D-MJSR) method. First, filter banks transferred from the pretrained VGG16 network are utilized to extract multiscale features of HSI. Then, features from each scale layer are respectively and collaboratively fused with the raw spectral feature and then upsampled by bilinear interpolation to reach the input size. Finally, with the advantages of feature distribution at different scales, JSR under multitask dictionaries is introduced to achieve the final classification, where samples are represented by dictionaries with different scale spaces independently, and neighborhood samples in each scale are represented by the same atoms wherever possible. We evaluate the proposed method D-MJSR by two public hyperspectral data sets quantitatively. Compared with existing feature extraction and SR-based methods, our method presents some significant improvement in classification accuracy.

Most of the existing spatial-spectral-based hyperspectral image classification (HSIC) methods mainly extract the spatial-spectral information by combining the pixels in a small neighborhood or aggregating the statistical and morphological characteristics. However, those strategies can only generate shallow appearance features with limited representative ability for classes with high interclass similarity and spatial diversity and therefore reduce the classification accuracy. To this end, we present a novel HSIC framework, named deep multiscale spatial-spectral feature extraction algorithm, which focuses on learning effective discriminant features for HSIC. First, the well pretrained deep fully convolutional network based on VGG-verydeep-16 is introduced to excavate the potential deep multiscale spatial structural information in the proposed hyperspectral imaging framework. Then, the spectral feature and the deep multiscale spatial feature are fused by adopting the weighted fusion method. Finally, the fusion feature is put into a generic classifier to obtain the pixelwise classification. Compared with the existing spectral-spatial-based classification techniques, the proposed method provides the state-of-the-art performance and is much more effective, especially for images with high nonlinear distribution and spatial diversity.