The use of low-dose computed tomography (LDCT) in medical practice can effectively reduce the radiation risk patients, but it may increase noise and artefacts, which compromise diagnostic information. methods based on deep learning improve image quality, most them a training set aligned pairs, are difficult to obtain practice. In order solve this problem, basis Wasserstein generative adversarial network (GAN) framework, we propose combining multi-perceptual loss fidelity loss....

Generative adversarial networks (GANs) have achieved many excellent results in hyperspectral image (HSI) classification recent years, as GANs can effectively solve the dilemma of limited training samples HSI classification. However, due to class imbalance problem data, always associate minority-class with fake label. To address this issue, we first propose a semi-supervised generative network incorporating transformer, called HyperViTGAN. The proposed HyperViTGAN is designed an external...

Hyperspectral image (HSI) classification attempts to classify each pixel, which is an important means of obtaining land–cover knowledge. images are cubic data with spectral–spatial knowledge and can generally be considered as sequential alongside spectral dimension. Unlike convolutional neural networks (CNNs), mainly focus on local relationship models in images, transformers have been shown a powerful structure for qualifying sequence data. However, it lacks the excellent ability CNNs...

Semi-supervised learning (SSL) focuses on the way to improve efficiency through use of labeled and unlabeled samples concurrently. However, recent research indicates that classification performance might be deteriorated by samples. Here, we proposed a novel graph-based semi-supervised algorithm combined with particle cooperation competition, which can model effectively using First, for purpose reducing generation label noise, used an efficient constrained graph construction approach...

Hyperspectral Images (HSIs) contain enriched information due to the presence of various bands, which have gained attention for past few decades. However, explosive growth in HSIs’ scale and dimensions causes “Curse dimensionality” “Hughes phenomenon”. Dimensionality reduction has become an important means overcome dimensionality”. In hyperspectral images, labeled samples are more difficult collect because they require many labor material resources. Semi-supervised dimensionality is very...

Recently, deep learning for hyperspectral image classification has been successfully applied, and some convolutional neural network (CNN)-based models already achieved attractive results. Since data is a spectral-spatial cube that can generally be considered as sequential along with the spectral dimension, CNN perform poorly on such data. Unlike networks (CNNs) mainly concern local relationship in images, transformer shown to powerful structure qualifying In SA (self-attention) module of...

Classification of Hyperspectral Images (HSIs) has gained attention for the past few decades. In remote sensing image classification, labeled samples are insufficient or hard to obtain; however, unlabeled ones frequently rich and a vast number. When there no sufficient samples, overfitting may occur. To resolve issue, in this present work, we proposed novel approach HSI feature extraction, called robust regularized Block Low-Rank Discriminant Analysis (BLRDA), which is efficient extraction...

Strip steel surface defect recognition is a pattern problem with wide applications. Previous works on strip mainly focus feature selection and dimension reduction. There are also approaches real-time systems that exploit the autocorrection within some given picture. However, instances cannot be used in practical applications because of bad rate low efficiency. In this paper, we study intelligent algorithm recognition, where goal to improve accuracy save running time. This very important...

Transformer has achieved outstanding performance in many fields such as computer vision benefiting from its powerful and efficient modelling ability long-range feature extraction capability complementary to convolution. However, on the one hand, lack of CNN's innate inductive biases, translation invariance local sensitivity, makes require more data for learning. On other labelled hyperspectral samples are scarce due time-consuming costly annotation task. To this end, we propose a...

Semisupervised Discriminant Analysis (SDA) is a semisupervised dimensionality reduction algorithm, which can easily resolve the out-of-sample problem. Relative works usually focus on geometric relationships of data points, are not obvious, to enhance performance SDA. Different from these relative works, regularized graph construction researched here, important in graph-based learning methods. In this paper, we propose novel for Analysis, called combined low-rank and <mml:math...

Unlike Support Vector Machine (SVM), Multiple Kernel Learning (MKL) allows datasets to be free choose the useful kernels based on their distribution characteristics rather than a precise one. It has been shown in literature that MKL holds superior recognition accuracy compared with SVM, however, at expense of time consuming computations. This creates analytical and computational difficulties solving algorithms. To overcome this issue, we first develop novel kernel approximation approach for...

Semisupervised Discriminant Analysis (SDA) aims at dimensionality reduction with both limited labeled data and copious unlabeled data, but it may fail to discover the intrinsic geometry structure when manifold is highly nonlinear. The kernel trick widely used map original nonlinearly separable problem an intrinsically larger space where classes are linearly separable. Inspired by low-rank representation (LLR), we proposed a novel SDA method called kernel-based (LRKSDA) algorithm LRR as...

The rapid dissemination of unverified information through social platforms like Twitter poses considerable dangers to societal stability. Identifying real versus fake claims is challenging, and previous work on rumor detection methods often fails effectively capture propagation structure features. These also overlook the presence comments irrelevant discussion topic source post. To address this, we introduce a novel approach: Structure-Aware Multilevel Graph Attention Network (SAMGAT) for...

Data integrity is a prerequisite for ensuring data availability of IoT and has received extensive attention in the field big security. Stream computing systems are widely used real-time acquisition computing. However, real-time, volatility, suddenness, disorder stream make verification difficult. According to survey, there no mature universal solution. To solve this issue, we constructed algorithm scheme system (S-DIV) by utilizing homomorphic message authentication code pseudo-random...

Nonnegative matrix factorization (NMF) model has been successfully applied to discover latent community structures due its good performance and interpretability advantages in extracting hidden patterns. However, most previous studies explore only the structural information of network while ignoring rich attributes. Besides, they aim at detecting densely connected communities (also called structures) fail identify general structures, such as bipartite mixture structures. In this paper, we...