检索结果-南通市图书馆

GPS: graph contrastive learning via multi-scale augmented views from adversarial pooling

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

science china(Information sciences) 2025年第1期68卷 145-158页

作者： Wei JU Yiyang GU Zhengyang MAO Ziyue QIAO Yifang QIN Xiao LUO Hui XIONG Ming ZHANG School of Computer science National Key Laboratory for Multimedia Information ProcessingPeking University Artificial Intelligence Thrust The Hong Kong University of Science and Technology Department of Computer science University of California

Self-supervised graph representation learning has recently shown considerable promise in a range of fields, including bioinformatics and social networks. A large number of graph contrastive learning approaches have shown promising performance for representation learning on graphs, which train models by maximizing agreement between original graphs and their augmented views(i.e., positive views). Unfortunately, these methods usually involve pre-defined augmentation strategies based on the knowledge of human experts. Moreover, these strategies may fail to generate challenging positive views to provide sufficient supervision signals. In this paper, we present a novel approach named graph pooling contrast(GPS) to address these *** by the fact that graph pooling can adaptively coarsen the graph with the removal of redundancy, we rethink graph pooling and leverage it to automatically generate multi-scale positive views with varying emphasis on providing challenging positives and preserving semantics, i.e., strongly-augmented view and weakly-augmented view. Then, we incorporate both views into a joint contrastive learning framework with similarity learning and consistency learning, where our pooling module is adversarially trained with respect to the encoder for adversarial robustness. Experiments on twelve datasets on both graph classification and transfer learning tasks verify the superiority of the proposed method over its counterparts.

关键词： graph representation learning graph neural networks graph contrastive learning graph augmentations graph pooling

Hypidone hydrochloride(YL-0919)ameliorates functional deficits after traumatic brain injury in mice by activating the sigma-1 receptor for antioxidation

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Neural Regeneration Research 2025年第8期20卷 2325-2336页

作者： Yafan Bai Hui Ma Yue Zhang Jinfeng Li Xiaojuan Hou Yixin Yang Guyan Wang Yunfeng Li Department of Anesthesiology Beijing Tongren HospitalCapital Medical UniversityBeijingChina Beijing Institute of Basic Medical sciences BeijingChina Hebei North University ZhangjiakouHebei ProvinceChina Beijing Institute of Pharmacology and Toxicology State Key Laboratory of Toxicology and Medical CountermeasuresBeijing Key Laboratory of NeuropsychopharmacologyBeijingChina

Traumatic brain injury involves complex pathophysiological mechanisms,among which oxidative stress significantly contributes to the occurrence of secondary *** this study,we evaluated hypidone hydrochloride(YL-0919),a self-developed antidepressant with selective sigma-1 receptor agonist properties,and its associated mechanisms and targets in traumatic brain *** experiments to assess functional deficits were followed by assessment of neuronal damage through histological analyses and examination of blood-brain barrier permeability and brain ***,we investigated the antioxidative effects of YL-0919 by assessing the levels of traditional markers of oxidative stress in vivo in mice and in vitro in HT22 ***,the targeted action of YL-0919 was verified by employing a sigma-1 receptor antagonist(BD-1047).Our findings demonstrated that YL-0919 markedly improved deficits in motor function and spatial cognition on day 3 post traumatic brain injury,while also decreasing neuronal mortality and reversing blood-brain barrier disruption and brain ***,YL-0919 effectively combated oxidative stress both in vivo and in *** protective effects of YL-0919 were partially inhibited by *** results indicated that YL-0919 relieved impairments in motor and spatial cognition by restraining oxidative stress,a neuroprotective effect that was partially reversed by the sigma-1 receptor antagonist ***-0919 may have potential as a new treatment for traumatic brain injury.

关键词： antidepressant drug blood-brain barrier cognitive function hypidone hydrochloride(YL-0919) neurological function nuclear factor-erythroid 2 related factor 2 oxidative stress sigma-1 receptor superoxide dismutase traumatic brain injury

维普期刊数据库

Personalized federated few-shot node classification

在线全文

学校读者我要写书评

暂无评论

science china(Information sciences) 2025年第1期68卷 204-218页

作者： Yunfeng ZHAO Xintong HE Guoxian YU Jun WANG Yongqing ZHENG Carlotta DOMENICONI School of Software Shandong University Joint SDU-NTU Centre for Artificial Intelligence Research Shandong University Department of Mathematics National University of Singapore Department of Computer science George Mason University

Personalized federated learning(PFL) aims to train customized models for individual clients in a decentralized setting, with the account of non-independent and identically distributed data across clients. However, most PFL methods adopt uniform classification layers for diverse clients and give rise to error-prone predictions, due to the task heterogeneity notably prominent in decentralized graph data scenarios. Although some PFL solutions setup client-specific classification layers for each client and optimize them only locally, they are corrupted with limited local training data. We propose an innovative solution called federated parameter decoupling and node augmentation(Fed PANo) to address these problems and to achieve personalized federated few-shot node classification, which is a prevalent and challenging but unexplored topic. Specifically, Fed PANo first separates the local model into the GNN and classifier to handle unique client-specific task variations. The GNN is trained through federated learning to capture shared knowledge of graph nodes across clients, while the classifier is custom-designed and trained individually for each client. Additionally, a generic classifier shared among clients is adopted to encourage the GNN's grasp of shared information. Then Fed PANo further proposes the node generator along with its local and collaborative training strategies to deal with the node scarcity of clients. Extensive experimental results on benchmark datasets confirm that Fed PANo outperforms eight competitive baselines across different settings.

关键词： personalized federated learning few-shot learning node classification task heterogeneity parameter decoupling node augmentation

Drone-YOLOv7: small target detection in UAV remote sensing images using YOLOv7-tiny with dynamic head

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Optoelectronics Letters 2025年

作者： Yang Ye Yang Sen Tong Jigang Wang Zenghui Tianjin University of Technology Department of Electrical Engineering， University of South Africa

In the task of detecting targets in unmanned aerial vehicles （UAVs） aerial images， the challenge arises from small size and wide scale span of targets， making their location， extraction and feature distinction difficult. Similar categories are susceptible to confusion. To address these issues， an improved YOLOv7-tiny aerial detection algorithm is proposed. Firstly， to improve the identification of small targets， a small target detection layer is added while removing the maximum target detection layer to enhance the localization ability of small targets and make the model lightweight. Secondly， a CBAM （Convolutional Block Attention Module） attention mechanism is incorporated at the connection between the backbone and neck to focus on the target area and eliminate interference. Additionally， to further enhance the classification and location capabilities of the targets， the detection head is replaced with a dynamic head （DYHead） with a unified channel number to fuse scale， space and task awareness in a single module. Finally， to overcome the difficulty of extracting features， the neck is replaced with a CS-FPN （Cross-Scale Feature Pyramid Network） structure with cross-scale connections to amplify feature fusion. Comparative experiments on the VisDrone2021 dataset show that the proposed model can improve mAP50and mAP0.5:0.95by 8.3% and 6.1%， respectively， compared to the original YOLOv7-tiny model， and significant improvements are also observed on CARPK and SeaDronesSee datasets.

关键词：

AsyCo: an asymmetric dual-task co-training model for partial-label learning

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

science china(Information sciences) 2025年

作者： Beibei LI Yiyuan ZHENG Beihong JIN Tao XIANG Haobo WANG Lei FENG College of Computer science Chongqing University State Key Laboratory of Computer science Institute of Software Chinese Academy of Sciences University of Chinese Academy of sciences School of Software Technology Zhejiang University School of Computer science and Engineering Nanyang Technological University

Partial-label learning (PLL) is a typical problem of weakly supervised learning, where each training instance is annotated with a set of candidate labels. Self-training PLL models achieve state-of-the-art performance but suffer from error accumulation problems caused by mistakenly disambiguated instances. Although co-training can alleviate this issue by training two networks simultaneously and allowing them to interact with each other, most existing co-training methods train two structurally identical networks with the same task, i.e., are symmetric, rendering it insufficient for them to correct each other due to their similar limitations. Therefore, in this paper, we propose an asymmetric dual-task co-training PLL model called AsyCo,which forces its two networks, i.e., a disambiguation network and an auxiliary network, to learn from different views explicitly by optimizing distinct tasks. Specifically, the disambiguation network is trained with a self-training PLL task to learn label confidence, while the auxiliary network is trained in a supervised learning paradigm to learn from the noisy pairwise similarity labels that are constructed according to the learned label confidence. Finally, the error accumulation problem is mitigated via information distillation and confidence refinement. Extensive experiments on both uniform and instance-dependent partially labeled datasets demonstrate the effectiveness of AsyCo.

关键词：

PVIIAD: industrial visual anomaly detection method based on partially visible image inpainting

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Optoelectronics Letters 2025年

作者： WU Qiliang FEI Dongce YAO Minghui NIU Yan WANG Cong School of Aeronautics and Astronautics， Tiangong University School of Artificial Intelligence， Tiangong University

Industrial anomaly detection is dedicated to identifying and locating regions that deviate from the standard appearance. The prevailing approach achieves unsupervised anomaly detection through the reconstruction of images using autoencoders. Due to the simplistic structure of some abnormal regions， the autoencoder can effectively reconstruct these areas， consequently diminishing the model's anomaly detection capabilities. To address this issue， this paper transforms the reconstruction task into the inpainting-filling-reconstruction task to increase the reconstruction error between abnormal samples and normal samples. In PVIIAD， the masked regions inpainted by the filling network are used to fill in the input image， thereby achieving an effect similar to masking. Unlike typical masking processes， this approach retains partial authentic information in the input image， rendering it partially visible. This is beneficial for the reconstruction network to repair the masked areas. Due to the consistent structure between the masked region inpainted by the filling network and the normal region， the filled abnormal regions display a complex structure that has not been learned， making it difficult for the reconstruction network to reconstruct the abnormal regions. Experimental results indicate that PVIIAD performs better than other methods on both the MVTec AD dataset and the MVTec LOCO AD dataset.

关键词：

Rendering acceleration method based on JND and sample gradient

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Optoelectronics Letters 2025年第3期21卷 177-182页

作者： ZHANG Ripei CHEN Chunyi School of Computer science and Technology Qinghai Normal University School of Computer science and Technology Changchun University of Science and Technology

Currently, the main idea of iterative rendering methods is to allocate a fixed number of samples to pixels that have not been fully rendered by calculating the completion rate. It is obvious that this strategy ignores the changes in pixel values during the previous rendering process, which may result in additional iterative operations.

关键词：

Formation of Singularity for Full Compressible Magnetohydrodynamic Equations with Zero Resistivity in Two Dimensional Bounded Domains

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Acta Mathematicae Applicatae Sinica 2023年第4期39卷 990-1008页

作者： Xin ZHONG School of Mathematics and Statistics Southwest UniversityChongqing 400715China

We are concerned with singularity formation of strong solutions to the two-dimensional(2D)full compressible magnetohydrodynamic equations with zero resistivity in a bounded *** energy method and critical Sobolev inequalities of logarithmic type,we show that the strong solution exists globally if the temporal integral of the maximum norm of the deformation tensor is *** result is the same as Ponce’s criterion for 3D incompressible Euler *** particular,it is independent of the magnetic field and ***,the initial vacuum states are allowed.

关键词： full compressible magnetohydrodynamic equations zero resistivity formation of singularity

维普期刊数据库

Adaptive fault-tolerant attitude tracking control for spacecraft with time-varying inertia uncertainties

在线全文

学校读者我要写书评

暂无评论

Chinese Journal of Aeronautics 2019年第3期32卷 674-687页

作者： Qinglei HU Li XIAO Chenliang WANG School of Automation science and Electrical Engineering Beihang University Sino-French Engineer School Beihang University

This paper is devoted to adaptive attitude tracking control for rigid spacecraft in the presence of parametric uncertainties, actuator faults and external disturbance. Specifically, a dynamic model is established based on one-tank spacecraft, which explicitly takes into account changing Center of Mass(CM). Then, a control scheme is proposed to achieve attitude *** from explicitly considering the changing CM during the controller design process, the proposed scheme possesses good robustness to parametric uncertainties with less fuel ***, a fault-tolerant control algorithm is proposed to accommodate actuator faults with no need of knowing the actuators' fault information. Lyapunov-based analysis is provided and the closed-loop system stability is rigorously proved. Finally, numerical simulations are presented to illustrate the effectiveness of the proposed controllers.

关键词： Adaptive control Attitude tracking Changing center of mass Fault-tolerant Spacecraft

维普期刊数据库