检索结果-内蒙古大学图书馆

Partial-Tuning Based Mixed-Modal Prototypes for Few-Shot Classification

ieee TRANSACTIONS ON MULTIMEDIA 2024年 26卷 9175-9186页

作者： Su, Yuling Liu, Xueliang Zhao, Ye Hong, Richang Wang, Meng Hefei Univ Technol Sch Comp Sci & Informat Engn Hefei 230009 Peoples R China Inst Artificial Intelligence Hefei Comprehens Natl Sci Ctr Hefei 230088 Peoples R China

The significant success of machine learning models is mainly based on a large amount of data for training iterations, but this limits their generalization for few-shot data. Some existing models utilize the extensive visual and textual modal knowledge of vision-language pre-trained models (VLPs) to compensate for the data scarcity problem. However, they may suffer from a classification bias problem during the fusion of multi-modal information since that they focus on the inter-modal matching while neglecting intra-modal recognition for few-shot images. In this paper, we propose a novel few-shot model with mixed-modal prototypes by partial-tuning the VLPs for better information fusion. It aims to yield a high-quality class prototype representation by integrating the abundant multi-modal knowledge of VLPs and the specific-task information of low-shot visual data. Specifically, we introduce an image-text alignment module to ensure the consistency of the few-shot visual representation and the textual knowledge of VLPs at the feature space. A self-similar learning module is designed to excavate the local and detailed characters of specific class, which is crucial under the data scarcity. Additionally, to preserve the generalizable pre-trained knowledge in the maximum extent, we partial-tune the parameters of VLPs to adapt for the few-shot tasks. To sum up, we mix multi-modal information at the feature representation level instead of fusing multi-modal matching similarities, which effectively mitigates classification bias and ultimately enhances the model performance for few-shot data. The extensive experiments are conducted to evaluate the effectiveness of our model on 11 benchmark datasets and the results show its promising.

关键词： visualization Prototypes Task analysis data models Adaptation models Training Metalearning Few-shot learning multi-modal learning partial-tuning vision-language pre-trained models

来源：评论

学校读者我要写书评

暂无评论

EFFECTS analysis OF SAR data QUANTIZATION ON DEEP LEARNING-BASED TARGET DETECTION TASK

EFFECTS ANALYSIS OF SAR DATA QUANTIZATION ON DEEP LEARNING-B...

引用

ieee International Geoscience and Remote Sensing symposium (IGARSS)

作者： Wang, Haochuan Yang, Wei Li, Chunsheng Sun, Bing Zeng, Hongcheng Beihang Univ Sch Elect & Informat Engn Beijing Peoples R China

ISBN: (纸本)9798350360332;9798350360325

Since the dynamic range of synthetic aperture radar (SAR) data is extremely large, SAR data quantization is required for storage and display of SAR images. The quantization process may cause undesirable change in the characteristics of the targets on the images, making it challenging to effectively detect targets in deep learning-based target detection task. To address this problem, a multi-quantization-based detection method is proposed in this paper. First, the effect of different quantization method is analysed and a multi-quantization based data augmentation strategy is proposed. Second, the labeling of SAR targets is analysed in response to the inconsistency between target scattering characteristics and physical contour shapes and the issue of partial visiblity. Then, the multi-quantization-based detection method is proposed to obtain more stable and complete detection results. The experiments conducted on the AIR-SARShip dataset demonstrate the effectiveness of the proposed method.

关键词： SAR quantization deep learning target detection

来源：评论

学校读者我要写书评

暂无评论

VDL-Surrogate: A View-Dependent Latent-based Model for Parameter Space Exploration of Ensemble Simulations

引用

ieee Transactions on visualization and Computer Graphics 2023年第1期29卷 820-830页

作者： Shi, Neng Xu, Jiayi Li, Haoyu Guo, Hanqi Woodring, Jonathan Shen, Han-Wei Department of Computer Science and Engineering The Ohio State University ColumbusOH43210 United States Mathematics and Computer Science Division Argonne National Laboratory LemontIL60439 United States Los Alamos National Laboratory Los AlamosNM87544 United States

We propose VDL-Surrogate, a view-dependent neural-network-latent-based surrogate model for parameter space exploration of ensemble simulations that allows high-resolution visualizations and user-specified visual mappings. Surrogate-enabled parameter space exploration allows domain scientists to preview simulation results without having to run a large number of computationally costly simulations. Limited by computational resources, however, existing surrogate models may not produce previews with sufficient resolution for visualization and analysis. To improve the efficient use of computational resources and support high-resolution exploration, we perform ray casting from different viewpoints to collect samples and produce compact latent representations. This latent encoding process reduces the cost of surrogate model training while maintaining the output quality. In the model training stage, we select viewpoints to cover the whole viewing sphere and train corresponding VDL-Surrogate models for the selected viewpoints. In the model inference stage, we predict the latent representations at previously selected viewpoints and decode the latent representations to data space. For any given viewpoint, we make interpolations over decoded data at selected viewpoints and generate visualizations with user-specified visual mappings. We show the effectiveness and efficiency of VDL-Surrogate in cosmological and ocean simulations with quantitative and qualitative evaluations. Source code is publicly available at https://***/trainsn/VDL-Surrogate. © 2022 ieee.

关键词： data visualization

来源：评论

学校读者我要写书评

暂无评论

NEAR-REAL-TIME EARTH OBSERVATION data ASSIMILATION FOR OCEAN CURRENTS

NEAR-REAL-TIME EARTH OBSERVATION DATA ASSIMILATION FOR OCEAN...

引用

ieee International Geoscience and Remote Sensing symposium (IGARSS)

作者： Baig, Shahbaz Qazi, Waqas A. Mumtaz, Rafia Natl Univ Sci & Technol Islamabad Pakistan OM Offshore Monitoring Ltd Limassol Cyprus

ISBN: (纸本)9798350360332;9798350360325

This paper presents a robust approach to assimilate Earth Observation (EO) data into multi-model ensembles, specifically focusing on integrating High-Frequency (HF) Radar-derived ocean surface currents covering Galician coast of Spain. The methodology employs Ensemble Kalman filter (EnKF) approach to promptly and effectively adjust the estimates of ensemble models in comparison to the provided observations. However, degradation of results is observed after 12 hrs forecast, due to variations in the covariance matrix. The performance of this approach has been assessed through comparative analysis between the averaged ensembles and four specific ensembles, employing the Root Mean Square Error (RMSE) metric. Our findings indicate that the assimilation of EO data into large ensembles of ocean currents models yields superior results compared to employing a single model. To extend the system's operational duration we propose the implementation of an inflation/regularization factor.

关键词： Ensemble Kalman Filter data Assimilation HF Radar Ocean Currents

来源：评论

学校读者我要写书评

暂无评论

Multivariate data Exploration Through Coordinated Views

引用

ieee ACCESS 2022年 10卷 78788-78817页

作者： Cruz, Antonio Arrais, Joel P. Machado, Penousal Univ Coimbra Ctr Informat & Syst Dept Informat Engn P-3030290 Coimbra Portugal LASI Intelligent Syst Associate Lab P-3030290 Coimbra Portugal

Many fields of study still face the challenges inherent to the analysis of complex multidimensional datasets, such as the field of computational biology, whose research of infectious diseases must contend with large protein-protein interaction networks with thousands of genes that vary in expression values over time. In this paper, we explore the visualization of multivariate data through CroP, a data visualization tool with a coordinated multiple views framework where users can adapt the workspace to different problems through flexible panels. In particular, we focus on the visualization of relational and temporal data, the latter being represented through layouts that distort timelines to represent the fluctuations of values across complex datasets, creating visualizations that highlight significant events and patterns. Moreover, CroP provides various layouts and functionalities to not only highlight relationships between different variables, but also dig-down into discovered patterns in order to better understand their sources and their effects. These methods are demonstrated through multiple experiments with diverse multivariate datasets, with a focus on gene expression time-series datasets. In addition to a discussion of our results, we also validate CroP through model and interface tests performed with participants from both the fields of information visualization and computational biology.

关键词： data visualization data models Analytical models Layout visualization Crops Biological system modeling Computational biology data visualization human-computer interaction interactive systems time-series analysis

来源：评论

学校读者我要写书评

暂无评论

Flexible And Faithful data Insights Generation 26

Flexible And Faithful Data Insights Generation

引用

26th International symposium on Multimedia, ISM 2024

作者： Zhang, Wei Bursztyn, Victor Soares Adobe Inc. McLeanVA United States Adobe Research San JoseCA United States

ISBN: (纸本)9798331511111

With the ever-growing volume of data, corporate users want to understand their data more efficiently. In addition to data visualization, they like text-based insights (e.g., a summary of trends, seasonality, and anomalies in the data) so they can digest them easily and/or share them with other stakeholders. In this paper, we introduce an efficient framework that generates faithful and flexible insights that resemble human writing. It is a hybrid approach that takes advantage of both template-based summaries and the latest Generative AI technology. With a fine-tuned compact large Language Model (LLM) and a Gatekeeper, we can achieve a lower error rate than state-of-the-art LLMs, while using only a fraction of resources. We have released a service based on the proposed approach and received very positive feedback. © 2024 ieee.

关键词： data assimilation

来源：评论

学校读者我要写书评

暂无评论

TOWARD SCALABLE DAMAGE ASSESSMENT FOR RAPID DISASTER RESPONSE

TOWARD SCALABLE DAMAGE ASSESSMENT FOR RAPID DISASTER RESPONS...

引用

ieee International Geoscience and Remote Sensing symposium (IGARSS)

作者： Wieland, Marc Hertel, Victor Geiss, Christian Martinis, Sandro Lechner, Konstanze German Aerosp Ctr DLR German Remote Sensing Data Ctr DFD Oberpfaffenhofen Germany

ISBN: (纸本)9798350360332;9798350360325

Current research and development efforts at DLR's Center for Satellite Based Crisis Information (ZKI) focus on deploying automated image analysis methods as part of rapid mapping processing routines. The use of machine learning methods enables processing of large amounts of heterogeneous satellite, aerial and drone images at varying spatial scales and temporal frequencies. In this work, we introduce an automated and scalable image processing chain for rapid building damage assessment, optimize it for inference on different hardware and provide application examples from recent natural disasters. We show the scalability of the method from high-frequency live-mapping with drones on a laptop to large-scale processing of satellite and aerial images on a high-performance computing cluster.

关键词： Damage Assessment Rapid Response Siamese Convolutional Neural Network Drones Optical Satellite Images

来源：评论

学校读者我要写书评

暂无评论

An Online visualization Framework for Massive 3D LiDAR Point Cloud data 4

An Online Visualization Framework for Massive 3D LiDAR Point...

引用

4th International Conference on Electronic Information Engineering and Computer Technology, EIECT 2024

作者： Qian, Junfeng Wang, Meng Yang, Ruman Ye, Haoting Tang, Jianjun Guizhou Power Grid Co Ltd Guiyang China The Chinese University of Hong Kong Shenzhen Shenzhen China Co. Ltd Guangzhou China

ISBN: (纸本)9798331528850

The accumulated LiDAR point cloud data is exploding, and its efficient visualization is critical to subsequent point cloud analysis and application. However, due to the large scale of point cloud data, traditional professional point cloud processing software usually has high hardware requirements with low visualization efficiency. Aiming at the problems, this paper designs an online visualization framework for massive 3D LiDAR point cloud data, which allows users to request the resources remotely through the network and interactively visualize the point cloud online in a progressive way. Firstly, the framework proposes a partition method of point cloud data based on the balanced KD-tree and a multi-level storage management scheme to store the point cloud data on HBase. Secondly, an adaptive spatial aggregation query strategy and efficient data transmission protocol based on WebSocket are proposed to support progressive on-demand exchange of massive point cloud dataset. Finally, multiple rendering schemes are designed to realize online visualization of the point cloud. In this paper, the Dutch AHN3 point cloud data is selected for performance validation experiments. Compared with the traditional methods, the online visualization effect is stable and efficient in this paper, and can provide a good interaction experience. © 2024 ieee.

关键词： Cloud platforms

来源：评论

学校读者我要写书评

暂无评论

Research on Fast analysis Model of Power Big data Integrating 3D data visualization Scene 3

Research on Fast Analysis Model of Power Big Data Integratin...

引用

2023 ieee 3rd International Conference on data Science and Computer Application, ICDSCA 2023

作者： Qingrui, Lin Bixing, Lin Shuze, Gao Jiangping, Yang Chaowang, Wu State Grid Info-Telecom Great Power Science and Technology Co. Ltd. Fujian Fuzhou350000 China

ISBN: (纸本)9798350341546

After determining large-scale power data and spatial scenarios, a new analysis model is constructed for 3D visualization scenarios. In this model, big data mining and 3D visualization technology are organically integrated, information fusion and fuzzy tomographic analysis method are adopted to realize efficient mining of power big data, and 3D rendering algorithm based on attribute reduction is used to improve the real-time performance of power 3D scene rendering. The visual feature reconstruction technology is used to realize the visual reconstruction of the 3D scene of power big data in the distribution network. The experiment shows that the data visualization management method can realize the efficient mining of power big data and real-time rendering of 3D model. © 2023 ieee.

关键词： 3D Scene Rendering Information Fusion Power data Three-Dimensional data Visual Scene

来源：评论

学校读者我要写书评

暂无评论

CriPAV: Street-Level Crime Patterns analysis and visualization

引用

ieee TRANSACTIONS ON visualization AND COMPUTER GRAPHICS 2022年第12期28卷 4000-4015页

作者： Garcia-Zanabria, Germain Raimundo, Marcos M. Poco, Jorge Nery, Marcelo Batista Silva, Claudio T. Adorno, Sergio Nonato, Luis Gustavo ICMC USP BR-13566590 Sao Carlos SP Brazil Fundacao Getulio Vargas BR-88010001 Florianopolis SC Brazil RIDC FAPESP BR-05508010 Sao Paulo Brazil Inst Adv Studies Global Cities Program BR-05508010 Sao Paulo Brazil NYU New York NY 10013 USA NEV CEPID USP BR-05508900 Sao Paulo Brazil

Extracting and analyzing crime patterns in big cities is a challenging spatiotemporal problem. The hardness of the problem is linked to two main factors, the sparse nature of the crime activity and its spread in large spatial areas. Sparseness hampers most time series (crime time series) comparison methods from working properly, while the handling of large urban areas tends to render the computational costs of such methods impractical. Visualizing different patterns hidden in crime time series data is another issue in this context, mainly due to the number of patterns that can show up in the time series analysis. In this article, we present a new methodology to deal with the issues above, enabling the analysis of spatiotemporal crime patterns in a street-level of detail. Our approach is made up of two main components designed to handle the spatial sparsity and spreading of crimes in large areas of the city. The first component relies on a stochastic mechanism from which one can visually analyze probablexintensive crime hotspots. Such analysis reveals important patterns that can not be observed in the typical intensity-based hotspot visualization. The second component builds upon a deep learning mechanism to embed crime time series in Cartesian space. From the embedding, one can identify spatial locations where the crime time series have similar behavior. The two components have been integrated into a web-based analytical tool called CriPAV (Crime Pattern analysis and visualization), which enables global as well as a street-level view of crime patterns. Developed in close collaboration with domain experts, CriPAV has been validated through a set of case studies with real crime data in Sao Paulo - Brazil. The provided experiments and case studies reveal the effectiveness of CriPAV in identifying patterns such as locations where crimes are not intense but highly probable to occur as well as locations that are far apart from each other but bear similar crime patterns.

关键词： data visualization Time series analysis Urban areas Tools Task analysis Deep learning Visual analytics Crime data spatio-temporal data visual analytics crime hotspots stochastic matrix

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：