Discriminating Tensor Spectral Clustering for High-Dimension-Low-Sample-Size Data.

Tensor spectral clustering (TSC) is a recently proposed approach to robustly group data into underlying clusters. Unlike the traditional spectral clustering (SC), which merely uses pairwise similarities of data in an affinity matrix, TSC aims at exploring their multiwise similarities in an affinity tensor to achieve better performance. However, the performance of TSC highly relies on the design of multiwise similarities, and it remains unclear especially for high-dimension-low-sample-size (HDLSS) data. To this end, this article has proposed a discriminating TSC (DTSC) for HDLSS data. Specifically, DTSC uses the proposed discriminating affinity tensor that encodes the pair-to-pair similarities, which are particularly constructed by the anchor-based distance. HDLSS asymptotic analysis shows that the proposed affinity tensor can explicitly differentiate samples from different clusters when the feature dimension is large. This theoretical property allows DTSC to improve the clustering performance on HDLSS data. Experimental results on synthetic and benchmark datasets demonstrate the effectiveness and robustness of the proposed method in comparison to several baseline methods.

Broad Multitask Learning System With Group Sparse Regularization.

The broad learning system (BLS) featuring lightweight, incremental extension, and strong generalization capabilities has been successful in its applications. Despite these advantages, BLS struggles in multitask learning (MTL) scenarios with its limited ability to simultaneously unravel multiple complex tasks where existing BLS models cannot adequately capture and leverage essential information across tasks, decreasing their effectiveness and efficacy in MTL scenarios. To address these limitations, we proposed an innovative MTL framework explicitly designed for BLS, named group sparse regularization for broad multitask learning system using related task-wise (BMtLS-RG). This framework combines a task-related BLS learning mechanism with a group sparse optimization strategy, significantly boosting BLS's ability to generalize in MTL environments. The task-related learning component harnesses task correlations to enable shared learning and optimize parameters efficiently. Meanwhile, the group sparse optimization approach helps minimize the effects of irrelevant or noisy data, thus enhancing the robustness and stability of BLS in navigating complex learning scenarios. To address the varied requirements of MTL challenges, we presented two additional variants of BMtLS-RG: BMtLS-RG with sharing parameters of feature mapped nodes (BMtLS-RGf), which integrates a shared feature mapping layer, and BMtLS-RGf and enhanced nodes (BMtLS-RGfe), which further includes an enhanced node layer atop the shared feature mapping structure. These adaptations provide customized solutions tailored to the diverse landscape of MTL problems. We compared BMtLS-RG with state-of-the-art (SOTA) MTL and BLS algorithms through comprehensive experimental evaluation across multiple practical MTL and UCI datasets. BMtLS-RG outperformed SOTA methods in 97.81% of classification tasks and achieved optimal performance in 96.00% of regression tasks, demonstrating its superior accuracy and robustness. Furthermore, BMtLS-RG exhibited satisfactory training efficiency, outperforming existing MTL algorithms by 8.04-42.85 times.

Heterogeneous Prototype Learning From Contaminated Faces Across Domains via Disentangling Latent Factors.

This article studies an emerging practical problem called heterogeneous prototype learning (HPL). Unlike the conventional heterogeneous face synthesis (HFS) problem that focuses on precisely translating a face image from a source domain to another target one without removing facial variations, HPL aims at learning the variation-free prototype of an image in the target domain while preserving the identity characteristics. HPL is a compounded problem involving two cross-coupled subproblems, that is, domain transfer and prototype learning (PL), thus making most of the existing HFS methods that simply transfer the domain style of images unsuitable for HPL. To tackle HPL, we advocate disentangling the prototype and domain factors in their respective latent feature spaces and then replacing the source domain with the target one for generating a new heterogeneous prototype. In doing so, the two subproblems in HPL can be solved jointly in a unified manner. Based on this, we propose a disentangled HPL framework, dubbed DisHPL, which is composed of one encoder-decoder generator and two discriminators. The generator and discriminators play adversarial games such that the generator embeds contaminated images into a prototype feature space only capturing identity information and a domain-specific feature space, while generating realistic-looking heterogeneous prototypes. Experiments on various heterogeneous datasets with diverse variations validate the superiority of DisHPL.

Synthesis and Evaluation of a Novel PET Radioligand for Imaging Glutaminyl Cyclase Activity as a Biomarker for Detecting Alzheimer's Disease.

Several new lines of research have demonstrated that a significant number of amyloid-ß peptides found in Alzheimer's disease (AD) are truncated and undergo post-translational modification by glutaminyl cyclase (QC) at the N-terminal. Notably, QC's products of Abeta-pE3 and Abeta-pE11 have been active targets for investigational drug development. This work describes the design, synthesis, characterization, and in vivo validation of a novel PET radioligand, [18F]PB0822, for targeted imaging of QC. We report herein a simplified and robust chemistry for the synthesis of the standard compound, [19F]PB0822, and the corresponding [18F]PB0822 radioligand. The PET probe was developed with 99.9% radiochemical purity, a molar activity of 965 Ci.mmol-1, and an IC50 of 56.3 nM, comparable to those of the parent PQ912 inhibitor (62.5 nM). Noninvasive PET imaging showed that the probe is distributed in the brain 5 min after intravenous injection. Further, in vivo PET imaging with [18F]PB0822 revealed that AD 5XFAD mice harbor significantly higher QC activity than WT counterparts. The data also suggested that QC activity is found across different brain regions of the tested animals.

Cross-Modal Hashing Method with Properties of Hamming Space: A New Perspective.

Cross-modal hashing (CMH) has attracted considerable attention in recent years. Almost all existing CMH methods primarily focus on reducing the modality gap and semantic gap, i.e., aligning multi-modal features and their semantics in Hamming space, without taking into account the space gap, i.e., difference between the real number space and the Hamming space. In fact, the space gap can affect the performance of CMH methods. In this paper, we analyze and demonstrate how the space gap affects the existing CMH methods, which therefore raises two problems: solution space compression and loss function oscillation. These two problems eventually cause the retrieval performance deteriorating. Based on these findings, we propose a novel algorithm, namely Semantic Channel Hashing (SCH). Firstly, we classify sample pairs into fully semantic-similar, partially semantic-similar, and semantic-negative ones based on their similarity and impose different constraints on them, respectively, to ensure that the entire Hamming space is utilized. Then, we introduce a semantic channel to alleviate the issue of loss function oscillation. Experimental results on three public datasets demonstrate that SCH outperforms the state-of-the-art methods. Furthermore, experimental validations are provided to substantiate the conjectures regarding solution space compression and loss function oscillation, offering visual evidence of their impact on the CMH methods. Codes are available at https://github.com/hutt94/SCH.

Multiview Tensor Spectral Clustering via Co-regularization.

Graph-based multi-view clustering encodes multi-view data into sample affinities to find consensus representation, effectively overcoming heterogeneity across different views. However, traditional affinity measures tend to collapse as the feature dimension expands, posing challenges in estimating a unified alignment that reveals both crossview and inner relationships. To tackle this challenge, we propose to achieve multi-view uniform clustering via consensus representation coregularization. First, the sample affinities are encoded by both popular dyadic affinity and recent high-order affinities to comprehensively characterize spatial distributions of the HDLSS data. Second, a fused consensus representation is learned through aligning the multi-view lowdimensional representation by co-regularization. The learning of the fused representation is modeled by a high-order eigenvalue problem within manifold space to preserve the intrinsic connections and complementary correlations of original data. A numerical scheme via manifold minimization is designed to solve the high-order eigenvalue problem efficaciously. Experiments on eight HDLSS datasets demonstrate the effectiveness of our proposed method in comparison with the recent thirteen benchmark methods.

Deep Tensor Spectral Clustering Network via Ensemble of Multiple Affinity Tensors.

Tensor spectral clustering (TSC) is an emerging approach that explores multi-wise similarities to boost learning. However, two key challenges have yet to be well addressed in the existing TSC methods: (1) The construction and storage of high-order affinity tensors to encode the multi-wise similarities are memory-intensive and hampers their applicability, and (2) they mostly employ a two-stage approach that integrates multiple affinity tensors of different orders to learn a consensus tensor spectral embedding, thus often leading to a suboptimal clustering result. To this end, this paper proposes a tensor spectral clustering network (TSC-Net) to achieve one-stage learning of a consensus tensor spectral embedding, while reducing the memory cost. TSC-Net employs a deep neural network that learns to map the input samples to the consensus tensor spectral embedding, guided by a TSC objective with multiple affinity tensors. It uses stochastic optimization to calculate a small part of the affinity tensors, thereby avoiding loading the whole affinity tensors for computation, thus significantly reducing the memory cost. Through using an ensemble of multiple affinity tensors, the TSC can dramatically improve clustering performance. Empirical studies on benchmark datasets demonstrate that TSC-Net outperforms the recent baseline methods.

Partial Multilabel Learning Using Noise-Tolerant Broad Learning System With Label Enhancement and Dimensionality Reduction.

Partial multilabel learning (PML) addresses the issue of noisy supervision, which contains an overcomplete set of candidate labels for each instance with only a valid subset of training data. Using label enhancement techniques, researchers have computed the probability of a label being ground truth. However, enhancing labels in the noisy label space makes it impossible for the existing partial multilabel label enhancement methods to achieve satisfactory results. Besides, few methods simultaneously involve the ambiguity problem, the feature space's redundancy, and the model's efficiency in PML. To address these issues, this article presents a novel joint partial multilabel framework using broad learning systems (namely BLS-PML) with three innovative mechanisms: 1) a trustworthy label space is reconstructed through a novel label enhancement method to avoid the bias caused by noisy labels; 2) a low-dimensional feature space is obtained by a confidence-based dimensionality reduction method to reduce the effect of redundancy in the feature space; and 3) a noise-tolerant BLS is proposed by adding a dimensionality reduction layer and a trustworthy label layer to deal with PML problem. We evaluated it on six real-world and seven synthetic datasets, using eight state-of-the-art partial multilabel algorithms as baselines and six evaluation metrics. Out of 144 experimental scenarios, our method significantly outperforms the baselines by about 80%, demonstrating its robustness and effectiveness in handling partial multilabel tasks.

Fully automated dual-run manufacturing of [11C]PIB on FASTlab.

This report describes an updated, fully automated method for the production of [11C]PIB on a cassette-based automated synthesis module. The method allows for two separate productions of [11C]PIB, both of which meet all specification for use in clinical studies. The GE FASTlab developer system was used to create the cassette design as well as the controlling tracer package. The method takes 16 min from the delivery of [11C]MeOTf to the FASTlab, or 35 min from the End of Bombardment; and reliably produces 3547 ± 586 MBq of [11C]PIB in high radiochemical purity (> 98 %). This methodology increases the production capacity of radiopharmaceutical facilities for [11C]PIB, and can easily produce 4 batches in a single day with limited infrastructure footprint.

Realize Generative Yet Complete Latent Representation for Incomplete Multi-View Learning.

In multi-view environment, it would yield missing observations due to the limitation of the observation process. The most current representation learning methods struggle to explore complete information by lacking either cross-generative via simply filling in missing view data, or solidative via inferring a consistent representation among the existing views. To address this problem, we propose a deep generative model to learn a complete generative latent representation, namely Complete Multi-view Variational Auto-Encoders (CMVAE), which models the generation of the multiple views from a complete latent variable represented by a mixture of Gaussian distributions. Thus, the missing view can be fully characterized by the latent variables and is resolved by estimating its posterior distribution. Accordingly, a novel variational lower bound is introduced to integrate view-invariant information into posterior inference to enhance the solidative of the learned latent representation. The intrinsic correlations between views are mined to seek cross-view generality, and information leading to missing views is fused by view weights to reach solidity. Benchmark experimental results in clustering, classification, and cross-view image generation tasks demonstrate the superiority of CMVAE, while time complexity and parameter sensitivity analyses illustrate the efficiency and robustness. Additionally, application to bioinformatics data exemplifies its practical significance.

Fully automated radiosynthesis of [68Ga]Ga-FAPI-46 with cyclotron produced gallium.

BACKGROUND: Radiopharmaceuticals capable of targeting the fibroblast activation protein have become widely utilized in the research realm as well as show great promise to be commercialized; with [68Ga]Ga-FAPI-46 being one of the most widely utilized. Until now the synthesis has relied on generator-produced gallium-68. Here we present a developed method to utilize liquid-target cyclotron-produced gallium-68 to prepare [68Ga]Ga-FAPI-46. RESULTS: A fully-automated manufacturing process for [68Ga]Ga-FAPI-46 was developed starting with the 68Zn[p,n]68Ga cyclotron bombardment to provide [68Ga]GaCl3, automated purification of the [68Ga]GaCl3, chelation with the precursor, and final formulation/purification. The activity levels produced were sufficient for multiple clinical research doses, and the final product met all release criteria. Furthermore, the process consistently provides < 2% of Ga-66 and Ga-67 at the 4-h expiry, meeting the Ph. Eur. CONCLUSIONS: The automated radiosynthesis on the GE FASTlab 2 module purifies the cyclotron output into [68Ga]GaCl3, performs the labeling, formulates the product, and sterilizes the product while transferring to the final vial. Production of > 40 mCi (> 1480 MBq) of [68Ga]Ga-FAPI-46 in excellent radiochemical yield was achieved with all batches meeting release criteria.

Abnormal developmental trajectory and vulnerability to cardiac arrhythmias in tetralogy of Fallot with DiGeorge syndrome.

Tetralogy of Fallot (TOF) is the most common cyanotic congenital heart disease. Ventricular dysfunction and cardiac arrhythmias are well-documented complications in patients with repaired TOF. Whether intrinsic abnormalities exist in TOF cardiomyocytes is unknown. We establish human induced pluripotent stem cells (hiPSCs) from TOF patients with and without DiGeorge (DG) syndrome, the latter being the most commonly associated syndromal association of TOF. TOF-DG hiPSC-derived cardiomyocytes (hiPSC-CMs) show impaired ventricular specification, downregulated cardiac gene expression and upregulated neural gene expression. Transcriptomic profiling of the in vitro cardiac progenitors reveals early bifurcation, as marked by ectopic RGS13 expression, in the trajectory of TOF-DG-hiPSC cardiac differentiation. Functional assessments further reveal increased arrhythmogenicity in TOF-DG-hiPSC-CMs. These findings are found only in the TOF-DG but not TOF-with no DG (ND) patient-derived hiPSC-CMs and cardiac progenitors (CPs), which have implications on the worse clinical outcomes of TOF-DG patients.

APLNR marks a cardiac progenitor derived with human induced pluripotent stem cells.

Cardiomyocytes can be readily derived from human induced pluripotent stem cell (hiPSC) lines, yet its efficacy varies across different batches of the same and different hiPSC lines. To unravel the inconsistencies of in vitro cardiac differentiation, we utilized single cell transcriptomics on hiPSCs undergoing cardiac differentiation and identified cardiac and extra-cardiac lineages throughout differentiation. We further identified APLNR as a surface marker for in vitro cardiac progenitors and immunomagnetically isolated them. Differentiation of isolated in vitro APLNR+ cardiac progenitors derived from multiple hiPSC lines resulted in predominantly cardiomyocytes accompanied with cardiac mesenchyme. Transcriptomic analysis of differentiating in vitro APLNR+ cardiac progenitors revealed transient expression of cardiac progenitor markers before further commitment into cardiomyocyte and cardiac mesenchyme. Analysis of in vivo human and mouse embryo single cell transcriptomic datasets have identified APLNR expression in early cardiac progenitors of multiple lineages. This platform enables generation of in vitro cardiac progenitors from multiple hiPSC lines without genetic manipulation, which has potential applications in studying cardiac development, disease modelling and cardiac regeneration.

Brain Network Classification for Accurate Detection of Alzheimer's Disease via Manifold Harmonic Discriminant Analysis.

Mounting evidence shows that Alzheimer's disease (AD) manifests the dysfunction of the brain network much earlier before the onset of clinical symptoms, making its early diagnosis possible. Current brain network analyses treat high-dimensional network data as a regular matrix or vector, which destroys the essential network topology, thereby seriously affecting diagnosis accuracy. In this context, harmonic waves provide a solid theoretical background for exploring brain network topology. However, the harmonic waves are originally intended to discover neurological disease propagation patterns in the brain, which makes it difficult to accommodate brain disease diagnosis with high heterogeneity. To address this challenge, this article proposes a network manifold harmonic discriminant analysis (MHDA) method for accurately detecting AD. Each brain network is regarded as an instance drawn on a Stiefel manifold. Every instance is represented by a set of orthonormal eigenvectors (i.e., harmonic waves) derived from its Laplacian matrix, which fully respects the topological structure of the brain network. An MHDA method within the Stiefel space is proposed to identify the group-dependent common harmonic waves, which can be used as group-specific references for downstream analyses. Extensive experiments are conducted to demonstrate the effectiveness of the proposed method in stratifying cognitively normal (CN) controls, mild cognitive impairment (MCI), and AD.

Remote Ischemic Conditioning in Pediatric Cancer Patients Receiving Anthracycline Chemotherapy: A Sham-Controlled Single-Blind Randomized Trial.

Background: Anthracycline cardiotoxicity is a concern in survivors of childhood cancers. Recent evidence suggests that remote ischemic conditioning (RIC) may offer myocardial protection. Objectives: This randomized sham-controlled single-blind study tested the hypothesis that RIC may reduce myocardial injury in pediatric cancer patients receiving anthracycline chemotherapy. Methods: We performed a phase 2 sham-controlled single-blind randomized controlled trial to determine the impact of RIC on myocardial injury in pediatric cancer patients receiving anthracycline-based chemotherapy. Patients were randomized to receive RIC (3 cycles of 5-minute inflation of a blood pressure cuff placed over 1 limb to 15 mm Hg above systolic pressure) or sham intervention. The intervention was applied within 60 minutes before initiation of the first dose and before up to 4 cycles of anthracycline therapy. The primary outcome was the plasma high-sensitivity cardiac troponin T (hs-cTnT) level. The secondary outcome measures included echocardiographic indexes of left ventricular systolic and diastolic function and the occurrence of cardiovascular events. Results: A total of 68 children 10.9 ± 3.9 years of age were randomized to receive RIC (n = 34) or sham (n = 34) intervention. Plasma levels of hs-cTnT showed a progressive increase across time points in the RIC (P < 0.001) and sham (P < 0.001) groups. At each of the time points, there were no significant differences in hs-cTnT levels or LV tissue Doppler and strain parameters between the 2 groups (all P > 0.05). None of the patients developed heart failure or cardiac arrhythmias. Conclusions: RIC did not exhibit cardioprotective effects in childhood cancer patients receiving anthracycline-based chemotherapy. (Remote Ischaemic Preconditioning in Childhood Cancer [RIPC]; NCT03166813).

Detection of QT interval prolongation using Apple Watch electrocardiogram in children and adolescents with congenital long QT syndrome.

Background: Apple watch-derived electrocardiogram (awECG) may help identify prolongation of corrected QT (QTc) interval. This study aimed to determine its usefulness for assessment of prolongation of QTc interval in children and adolescents with long QT syndrome (LQTS). Methods: Children and adolescents with and without LQTS were recruited for measurement of QTc intervals based on standard 12-lead (sECG) and awECG lead I, II and V5 tracings. Bland-Altman analysis of reproducibility, concordance assessment of T wave morphologies, and receiver operating characteristic (ROC) analysis of sensitivity and specificity of awECG-derived QTc interval for detecting QTc prolongation were performed. Results: Forty-nine patients, 19 with and 30 without LQTS, aged 3-22 years were studied. The intraclass correlation coefficient was 1.00 for both intra- and inter-observer variability in the measurement of QTc interval. The awECG- and sECG-derived QTc intervals correlated strongly in all three leads (r = 0.90-0.93, all p < 0.001). Concordance between awECG and sECG in assessing T wave morphologies was 84% (16/19). For detection of QTc prolongation, awECG lead V5 had the best specificity (94.4% and 87.5%, respectively) and positive predictive value (87.5% and 80.0%, respectively), and for identification of patients with LQTS, awECG leads II and V5 had the greatest specificity (92.3%-94.1%) and positive predictive value (85.7% to 91.7%) in both males and females. Conclusions: Apple Watch leads II and V5 tracings can be used for reproducible and accurate measurement of QTc interval, ascertainment of abnormal T wave morphologies, and detection of prolonged QTc interval in children and adolescents with LQTS.

Characterization of an Emergent Chicken H3N8 Influenza Virus in Southern China: a Potential Threat to Public Health.

Although influenza A viruses of several subtypes have occasionally infected humans, to date only those of the H1, H2, and H3 subtypes have led to pandemics and become established in humans. The detection of two human infections by avian H3N8 viruses in April and May of 2022 raised pandemic concerns. Recent studies have shown the H3N8 viruses were introduced into humans from poultry, although their genesis, prevalence, and transmissibility in mammals have not been fully elucidated. Findings generated from our systematic influenza surveillance showed that this H3N8 influenza virus was first detected in chickens in July 2021 and then disseminated and became established in chickens over wider regions of China. Phylogenetic analyses revealed that the H3 HA and N8 NA were derived from avian viruses prevalent in domestic ducks in the Guangxi-Guangdong region, while all internal genes were from enzootic poultry H9N2 viruses. The novel H3N8 viruses form independent lineages in the glycoprotein gene trees, but their internal genes are mixed with those of H9N2 viruses, indicating continuous gene exchange among these viruses. Experimental infection of ferrets with three chicken H3N8 viruses showed transmission through direct contact and inefficient transmission by airborne exposure. Examination of contemporary human sera detected only very limited antibody cross-reaction to these viruses. The continuing evolution of these viruses in poultry could pose an ongoing pandemic threat. IMPORTANCE A novel H3N8 virus with demonstrated zoonotic potential has emerged and disseminated in chickens in China. It was generated by reassortment between avian H3 and N8 virus(es) and long-term enzootic H9N2 viruses present in southern China. This H3N8 virus has maintained independent H3 and N8 gene lineages but continues to exchange internal genes with other H9N2 viruses to form novel variants. Our experimental studies showed that these H3N8 viruses were transmissible in ferrets, and serological data suggest that the human population lacks effective immunological protection against it. With its wide geographical distribution and continuing evolution in chickens, other spillovers to humans can be expected and might lead to more efficient transmission in humans.

Physiological concentration of protocatechuic acid directly protects vascular endothelial function against inflammation in diabetes through Akt/eNOS pathway.

Background: Cardiovascular diseases (CVDs) have been the major cause of mortality in type 2 diabetes. However, new approaches are still warranted since current diabetic medications, which focus mainly on glycemic control, do not effectively lower cardiovascular mortality rate in diabetic patients. Protocatechuic acid (PCA) is a phenolic acid widely distributed in garlic, onion, cauliflower and other plant-based foods. Given the anti-oxidative effects of PCA in vitro, we hypothesized that PCA would also have direct beneficial effects on endothelial function in addition to the systemic effects on vascular health demonstrated by previous studies. Methods and results: Since IL-1ß is the major pathological contributor to endothelial dysfunction in diabetes, the anti-inflammatory effects of PCA specific on endothelial cells were further verified by the use of IL-1ß-induced inflammation model. Direct incubation of db/db mouse aortas with physiological concentration of PCA significantly ameliorated endothelium-dependent relaxation impairment, as well as reactive oxygen species overproduction mediated by diabetes. In addition to the well-studied anti-oxidative activity, PCA demonstrated strong anti-inflammatory effects by suppressing the pro-inflammatory cytokines MCP1, VCAM1 and ICAM1, as well as increasing the phosphorylation of eNOS and Akt in the inflammatory endothelial cell model induced by the key player in diabetic endothelial dysfunction IL-1ß. Upon blocking of Akt phosphorylation, p-eNOS/eNOS remained low and the inhibition of pro-inflammatory cytokines by PCA ceased. Conclusion: PCA exerts protection on vascular endothelial function against inflammation through Akt/eNOS pathway, suggesting daily acquisition of PCA may be encouraged for diabetic patients.

SAGN: Semantic-Aware Graph Network for Remote Sensing Scene Classification.

The scene classification of remote sensing (RS) images plays an essential role in the RS community, aiming to assign the semantics to different RS scenes. With the increase of spatial resolution of RS images, high-resolution RS (HRRS) image scene classification becomes a challenging task because the contents within HRRS images are diverse in type, various in scale, and massive in volume. Recently, deep convolution neural networks (DCNNs) provide the promising results of the HRRS scene classification. Most of them regard HRRS scene classification tasks as single-label problems. In this way, the semantics represented by the manual annotation decide the final classification results directly. Although it is feasible, the various semantics hidden in HRRS images are ignored, thus resulting in inaccurate decision. To overcome this limitation, we propose a semantic-aware graph network (SAGN) for HRRS images. SAGN consists of a dense feature pyramid network (DFPN), an adaptive semantic analysis module (ASAM), a dynamic graph feature update module, and a scene decision module (SDM). Their function is to extract the multi-scale information, mine the various semantics, exploit the unstructured relations between diverse semantics, and make the decision for HRRS scenes, respectively. Instead of transforming single-label problems into multi-label issues, our SAGN elaborates the proper methods to make full use of diverse semantics hidden in HRRS images to accomplish scene classification tasks. The extensive experiments are conducted on three popular HRRS scene data sets. Experimental results show the effectiveness of the proposed SAGN.