检索结果-内蒙古大学图书馆

On learning the right attention point for feature enhancement

science China(Information sciences) 2023年第1期66卷 131-143页

作者： Liqiang LIN Pengdi HUANG Chi-Wing FU Kai XU Hao ZHANG Hui HUANG College of Computer Science and Software Engineering Shenzhen University Department of Computer Science and Engineering The Chinese University of Hong Kong School of Computer Science National University of Defense Technology School of Computing Science Simon Fraser University

We present a novel attention-based mechanism to learn enhanced point features for point cloud processing tasks, e.g., classification and segmentation. Unlike prior studies, which were trained to optimize the weights of a pre-selected set of attention points, our approach learns to locate the best attention points to maximize the performance of a specific task, e.g., point cloud classification. Importantly, we advocate the use of single attention point to facilitate semantic understanding in point feature learning. Specifically,we formulate a new and simple convolution, which combines convolutional features from an input point and its corresponding learned attention point(LAP). Our attention mechanism can be easily incorporated into state-of-the-art point cloud classification and segmentation networks. Extensive experiments on common benchmarks, such as Model Net40, Shape Net Part, and S3DIS, all demonstrate that our LAP-enabled networks consistently outperform the respective original networks, as well as other competitive alternatives, which employ multiple attention points, either pre-selected or learned under our LAP framework.

关键词： point convolution feature enhancement attention point deep neural network

来源：评论

学校读者我要写书评

暂无评论

Urban Impact on Landfalling Tropical Cyclone Precipitation:A Numerical Study of Typhoon Rumbia(2018)

引用

Advances in Atmospheric sciences 2023年第6期40卷 988-1004页

作者： Xinguan DU Haishan CHEN Qingqing LI Xuyang GE KLME/ILCEC/CIC-FEMD Nanjing University of Information Science and TechnologyNanjing 210044China School of Atmospheric Sciences Nanjing University of Information Science and TechnologyNanjing 210044China

Coastal urban areas are prone to serious disasters caused by landfalling tropical cyclones(TCs). Despite the crucial role of urban forcing in precipitation, how fine-scale urban features impact landfalling TC precipitation remains poorly understood. In this study, high-resolution ensemble simulations of Typhoon Rumbia(2018), which crossed the Yangtze River Delta urban agglomeration, were conducted to analyze the potential urban impact on TC precipitation. Results show that the inner-core rainfall of Rumbia is strengthened by approximately 10% due to the urban impact near the landfall,whereas minor differences in outer-core rainfall are found when the urban impact is excluded. Further diagnostic analyses indicate that low-level upward motion is crucial for precipitation evolution, as both co-vary during landfall. Moreover, the frictionally induced upward motion plays a decisive role in enhancing the rainfall when the urban impacts are *** surface friction can decelerate the tangential wind and therefore destroy the gradient balance and strengthen the radial wind within the boundary layer and thus can enhance upward motion. This study demonstrates that urban surface friction and related physical processes make the most significant contribution to landfalling TC rainfall enhancement.

关键词： landfalling tropical cyclone urban agglomeration quantitative diagnosis rainfall physical mechanisms surface friction

来源：评论

学校读者我要写书评

暂无评论

Combining Innovative CVTNet and Regularization Loss for Robust Adversarial Defense

引用

Journal of computer science & technology 2024年第5期39卷 1078-1093页

作者： Wei-Dong Wang Zhi Li Li Zhang Laboratory of Public Big Data School of Computer Science and TechnologyGuizhou UniversityGuiyang 550025China

Deep neural networks(DNNs)are vulnerable to elaborately crafted and imperceptible adversarial *** the continuous development of adversarial attack methods,existing defense algorithms can no longer defend against them ***,numerous studies have shown that vision transformer(ViT)has stronger robustness and generalization performance than the convolutional neural network(CNN)in various ***,because the standard denoiser is subject to the error amplification effect,the prediction network cannot correctly classify all reconstruction ***,this paper proposes a defense network(CVTNet)that combines CNNs and ViTs that is appended in front of the prediction *** can effectively eliminate adversarial perturbations and maintain high ***,this paper proposes a regularization loss(L_(CPL)),which optimizes the CVTNet by computing different losses for the correct prediction set(CPS)and the wrong prediction set(WPS)of the reconstruction examples,*** evaluation results on several standard benchmark datasets show that CVTNet performs better robustness than other advanced *** with state-of-the-art algorithms,the proposed CVTNet defense improves the average accuracy of pixel-constrained attack examples generated on the CIFAR-10 dataset by 24.25%and spatially-constrained attack examples by 14.06%.Moreover,CVTNet shows excellent generalizability in cross-model protection.

关键词： deep learning adversarial defense vision transformer image reconstruction

来源：评论

学校读者我要写书评

暂无评论

A Pre-Selection-Based Ant Colony System for Integrated Resources Scheduling Problem at Marine Container Terminal

引用

computers, Materials & Continua 2024年第8期80卷 2363-2385页

作者： Rong Wang Xinxin Xu Zijia Wang Fei Ji Nankun Mu School of Electronic and Information Engineering South China University of TechnologyGuangzhou510641China School of Computer Science and Technology Ocean University of ChinaQingdao266100China School of Computer Science and Cyber Engineering Guangzhou UniversityGuangzhou510006China School of Computer Science Chongqing UniversityChongqing400044China

Marine container terminal(MCT)plays a key role in the marine intelligent transportation system and international logistics ***,the efficiency of resource scheduling significantly influences the operation performance of *** solve the practical resource scheduling problem(RSP)in MCT efficiently,this paper has contributions to both the problem model and the algorithm ***,in the problem model,different from most of the existing studies that only consider scheduling part of the resources in MCT,we propose a unified mathematical model for formulating an integrated *** new integrated RSP model allocates and schedules multiple MCT resources simultaneously by taking the total cost minimization as the ***,in the algorithm design,a pre-selection-based ant colony system(PACS)approach is proposed based on graphic structure solution representation and a pre-selection *** the one hand,as the RSP can be formulated as the shortest path problem on the directed complete graph,the graphic structure is proposed to represent the solution encoding to consider multiple constraints and multiple factors of the RSP,which effectively avoids the generation of infeasible *** the other hand,the pre-selection strategy aims to reduce the computational burden of PACS and to fast obtain a higher-quality *** evaluate the performance of the proposed novel PACS in solving the new integrated RSP model,a set of test cases with different sizes is *** results and comparisons show the effectiveness and efficiency of the PACS algorithm,which can significantly outperform other state-of-the-art algorithms.

关键词： Resource scheduling problem(RSP) ant colony system(ACS) marine container terminal(MCT) pre-selection strategy

来源：评论

学校读者我要写书评

暂无评论

Hybrid tri-memristor hyperchaotic map and application in Wasserstein Generative Adversarial Nets

引用

science China(Technological sciences) 2024年第6期67卷 1855-1865页

作者： GU Yang BAO Han YU XiHong HUA ZhongYun BAO BoCheng XU Quan School of Microelectronics and Control Engineering Changzhou UniversityChangzhou213159China School of Computer Science and Technology Harbin Institute of TechnologyShenzhen518055China

Inspired by basic circuit connection methods,memristors can also be utilized in the construction of complex discrete chaotic *** investigate the dynamical effects of hybrid memristors,we propose two hybrid tri-memristor hyperchaotic(HTMH)mapping structures based on the hybrid parallel/cascade and cascade/parallel operations,*** the HTMH mapping structure with hybrid parallel/cascade operation as an example,this map possesses a spatial invariant set whose stability is closely related to the initial states of the *** distributions and bifurcation behaviours dependent on the control parameters are explored with numerical ***,the memristor initial offset-boosting mechanism is theoretically demonstrated,and memristor initial offset-boosting behaviours are numerically *** results clarify that the HTMH map can exhibit hyperchaotic behaviours and extreme multistability with homogeneous coexisting infinite *** addition,an FPGA hardware platform is fabricated to implement the HTMH map and generate pseudorandom numbers(PRNs)with high ***,the generated PRNs can be applied in Wasserstein generative adversarial nets(WGANs)to enhance training stability and generation capability.

关键词： memristor hybrid tri-memristor map spatial invariant set hyperchaos extreme multistability generative adversarial nets

来源：评论

学校读者我要写书评

暂无评论

Online Nonstop Task Management for Storm-Based Distributed Stream Processing Engines

引用

Journal of computer science & technology 2024年第1期39卷 116-138页

作者：张洲金培权谢希科王晓亮刘睿诚万寿红 School of Computer Science and Technology University of Science and Technology of ChinaHefei 230026China Key Laboratory of Electromagnetic Space Information Chinese Academy of SciencesHefei 230026China

Most distributed stream processing engines(DSPEs)do not support online task management and cannot adapt to time-varying data ***,some studies have proposed online task deployment algorithms to solve this ***,these approaches do not guarantee the Quality of Service(QoS)when the task deployment changes at runtime,because the task migrations caused by the change of task deployments will impose an exorbitant *** study one of the most popular DSPEs,Apache Storm,and find out that when a task needs to be migrated,Storm has to stop the resource(implemented as a process of Worker in Storm)where the task is *** will lead to the stop and restart of all tasks in the resource,resulting in the poor performance of task *** to solve this problem,in this pa-per,we propose N-Storm(Nonstop Storm),which is a task-resource decoupling DSPE.N-Storm allows tasks allocated to resources to be changed at runtime,which is implemented by a thread-level scheme for task ***,we add a local shared key/value store on each node to make resources aware of the changes in the allocation ***,each resource can manage its tasks at *** on N-Storm,we further propose Online Task Deployment(OTD).Differ-ing from traditional task deployment algorithms that deploy all tasks at once without considering the cost of task migra-tions caused by a task re-deployment,OTD can gradually adjust the current task deployment to an optimized one based on the communication cost and the runtime states of *** demonstrate that OTD can adapt to different kinds of applications including computation-and communication-intensive *** experimental results on a real DSPE cluster show that N-Storm can avoid the system stop and save up to 87%of the performance degradation time,compared with Apache Storm and other state-of-the-art *** addition,OTD can increase the average CPU usage by 51%for computation-intensive application

关键词： distributed stream processing engine(DSPE) Apache Storm online task migration online task deployment

来源：评论

学校读者我要写书评

暂无评论

A multilayer network diffusion-based model for reviewer recommendation

引用

Chinese Physics B 2024年第3期33卷 700-717页

作者：黄羿炜徐舒琪蔡世民吕琳媛 Institute of Fundamental and Frontier Sciences University of Electronic Science and Technology of ChinaChengdu 611731China Yangtze Delta Region Institute (Huzhou) University of Electronic Science and Technology of ChinaHuzhou 313001China Institute of Dataspace Hefei Comprehensive National Science CenterHefei 230088China Big Data Research Center University of Electronic Science and Technology of ChinaChengdu 611731China School of Computer Science and Engineering University of Electronic Science and Technology of ChinaChengdu 611731China School of Cyber Science and Technology University of Science and Technology of ChinaHefei 230026China

With the rapid growth of manuscript submissions,finding eligible reviewers for every submission has become a heavy *** systems are powerful tools developed in computer science and information science to deal with this ***,most existing approaches resort to text mining techniques to match manuscripts with potential reviewers,which require high-quality textual information to perform *** this paper,we propose a reviewer recommendation algorithm based on a network diffusion process on a scholar-paper multilayer network,with no requirement for textual *** network incorporates the relationship of scholar-paper pairs,the collaboration among scholars,and the bibliographic coupling among *** results show that our proposed algorithm outperforms other state-of-the-art recommendation methods that use graph random walk and matrix factorization and methods that use machine learning and natural language processing,with improvements of over 7.62%in recall,5.66%in hit rate,and 47.53%in ranking *** work sheds light on the effectiveness of multilayer network diffusion-based methods in the reviewer recommendation problem,which will help to facilitate the peer-review process and promote information retrieval research in other practical scenes.

关键词： reviewer recommendation multilayer network network diffusion model recommender systems complex networks

来源：评论

学校读者我要写书评

暂无评论

Enhanced Temporal Correlation for Universal Lesion Detection

引用

computer Modeling in Engineering & sciences 2024年第3期138卷 3051-3063页

作者： Muwei Jian Yue Jin Hui Yu School of Computer Science and Technology Shandong University of Finance and EconomicsJinanChina School of Information Science and Technology Linyi UniversityLinyiChina School of Creative Technologies University of PortsmouthPortsmouthUK

Universal lesion detection(ULD)methods for computed tomography(CT)images play a vital role in the modern clinical medicine and intelligent *** is well known that single 2D CT slices lack spatial-temporal characteristics and contextual information compared to 3D CT ***,3D CT blocks necessitate significantly higher hardware resources during the learning ***,efficiently exploiting temporal correlation and spatial-temporal features of 2D CT slices is crucial for ULD *** this paper,we propose a ULD network with the enhanced temporal correlation for this purpose,named *** designed TCE module is applied to enrich the discriminate feature representation of multiple sequential CT ***,we employ multi-scale feature maps to facilitate the localization and detection of lesions in various *** experiments are conducted on the DeepLesion benchmark demonstrate that thismethod achieves 66.84%and 78.18%for FS@0.5 and FS@1.0,respectively,outperforming compared state-of-the-art methods.

关键词： Universal lesion detection computational biology medical computing deep learning enhanced temporal correlation

来源：评论

学校读者我要写书评

暂无评论

Exposing Audio-Visual Forgeries in Frequency Domain 18th

Exposing Audio-Visual Forgeries in Frequency Domain

引用

18th Chinese Conference on Biometric Recognition, CCBR 2024

作者： Wan, Yuanfei Wang, Jian Cui, Jinrong Sun, Yunlian School of Computer Science and Engineering Nanjing University of Science and Technology Nanjing China School of Computer Science Nanjing University of Posts and Telecommunications Nanjing China South China Agricultural University Guangzhou China

ISBN: (纸本)9789819610679

Recently, the rapid development of deepfake technology attracted strong attention from the community. Some previous work on deepfake detection achieved good results in the frequency domain, which inspires us to combine frequency-domain information with temporal and spatial domains of visual to detect deepfakes. In addition, the audio signal can be represented in the frequency domain, so we can explore multimodal frequency-domain cues by combining audio and visual modalities. In this paper, we propose a Frequency-aware Audio-Visual Deepfake Detection(FAVDD) method. Specifically, we design a Frequency-Temporal-Spatial(FTS) visual encoder that extracts spatial, frequency, and temporal forgery cues and embeds them into visual features to form a unified representation. In addition, we project the audio signal into the frequency domain by Fourier transform and capture the forgery traces, which are later combined with visual features for deepfake detection. The results show that our proposed framework effectively combines multiple cues and achieves good results on three multimodal deepfake datasets. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Audio acoustics

来源：评论

学校读者我要写书评

暂无评论

LDTR:Transformer-based lane detection with anchor-chain representation

引用

Computational Visual Media 2024年第4期10卷 753-769页

作者： Zhongyu Yang Chen Shen Wei Shao Tengfei Xing Runbo Hu Pengfei Xu Hua Chai Ruini Xue School of Computer Science and Engineering University of Electronic Science and Technology of ChinaChengdu 611731China Didi Chuxing Beijing 100081China

Despite recent advances in lane detection methods,scenarios with limited-or no-visual-clue of lanes due to factors such as lighting conditions and occlusion remain challenging and crucial for automated ***,current lane representations require complex post-processing and struggle with specific *** by the DETR architecture,we propose LDTR,a transformer-based model to address these *** are modeled with a novel anchorchain,regarding a lane as a whole from the beginning,which enables LDTR to handle special lanes *** enhance lane instance perception,LDTR incorporates a novel multi-referenced deformable attention module to distribute attention around the ***,LDTR incorporates two line IoU algorithms to improve convergence efficiency and employs a Gaussian heatmap auxiliary branch to enhance model representation capability during *** evaluate lane detection models,we rely on Fr´echet distance,parameterized F1-score,and additional synthetic *** results demonstrate that LDTR achieves state-of-the-art performance on well-known datasets.

关键词： transformer lane detection anchor-chain

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：