检索结果-内蒙古大学图书馆

Proceedings of the 38th International Conference on Neural Information Processing Systems

作者： Xuexun Liu Xiaoxu Xu Jinlong Li Qiudan Zhang Xu Wang Nicu Sebe Lin Ma College of Computer Science and Software Engineering Shenzhen University Shenzhen China University of Trento Italy Meituan Inc. China

ISBN: (纸本)9798331314385

Referring 3D Segmentation is a visual-language task that segments all points of the specified object from a 3D point cloud described by a sentence of query. Previous works perform a two-stage paradigm, first conducting language-agnostic instance segmentation then matching with given text query. However, the semantic concepts from text query and visual cues are separately interacted during the training, and both instance and semantic labels for each object are required, which is time consuming and human-labor intensive. To mitigate these issues, we propose a novel Referring 3D Segmentation pipeline, Label-Efficient and Single-Stage, dubbed LESS, which is only under the supervision of efficient binary mask. Specifically, we design a Point-Word Cross-Modal Alignment module for aligning the fine-grained features of points and textual embedding. Query Mask Predictor module and Query-Sentence Alignment module are introduced for coarse-grained alignment between masks and query. Furthermore, we propose an area regularization loss, which coarsely reduces irrelevant background predictions on a large scale. Besides, a point-to-point contrastive loss is proposed concentrating on distinguishing points with subtly similar features. Through extensive experiments, we achieve state-of-the-art performance on ScanRefer dataset by surpassing the previous methods about 3.7% mIoU using only binary labels. Code is available at https://***/mellody11/LESS.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Exploiting Secure Multi-Hop Transmissions with NOMA Networks: Performance Analysis

Exploiting Secure Multi-Hop Transmissions with NOMA Networks...

引用

International Conference on Artificial Intelligence in Information and Communication (ICAIIC)

作者： Yushintia Pramitarini Ridho Hendra Yoga Perdana Kyusung Shim Beongku An Dept. of Software and Communications Engineering Graduate School Hongik University Republic of Korea School of Computer Engineering and Applied Mathematics Hankyong National University Republic of Korea Dept. of Software and Communications Engineering Hongik University Republic of Korea

In this paper, we analyze the secrecy performance of secure multi-hop transmission for the NOMA system. More specifically, the multi-hop transmission can extend the coverage of wireless transmission. However, according to the principle of the wireless medium, an eavesdropper can overhear the legitimate users' transmission, which leads to serious security issues. In order to analyze the relationship between network parameters and secrecy outage probability, we derive exact closed-form expressions for the secrecy outage probability (SOP) of cell center and cell edge users, respectively. The numerical results show that the simulation and analysis results are tightly matched. Additionally, we investigate the impact of the number of hops and the distance on the secrecy performance.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Research and Development of Kubernetes Operator for Machine Learning Pipelines

Research and Development of Kubernetes Operator for Machine ...

引用

International Russian Smart Industry Conference (SmartIndustryCon)

作者： Andrey Trapezin Ivan Filianin Faculty of Programming and Infocommunications ITMO University St. Petersburg Russia Faculty of Software Engineering and Computer Engineering ITMO University St. Petersburg Russia

ISBN: (数字)9798331511241

ISBN: (纸本)9798331511258

The current research work addresses the problem of automating the delivery of machine learning models from MLflow to Kubernetes infrastructure. To solve the mentioned problem, a Kubernetes operator has been developed to automate the delivery of machine learning models to production by integrating MLflow for model tracking and Seldon Core for model serving. The developed operator allows data scientists to deploy models while maintaining the familiar MLflow environment. The operator's automatic deployment triggers upon tagging models in MLflow, greatly simplifying engineers' tasks and minimizing the need for manual infrastructure configuration. By automating configuration tasks and optimizing deployment workflows, the solution achieves a 40-50% reduction in model time to deployment (TTD) metric compared to manual processes and decreases error rates from 15% to around 3%. The practical relevance of the work is that it simplifies collaboration between data and infrastructure teams by providing a unified deployment framework, resulting in faster, more reliable, and automated integration of machine learning models into an organisation's business processes.

关键词： Production systems Error analysis Machine learning Manuals Tagging Data models Reliability Research and development Load modeling Resilience

来源：评论

学校读者我要写书评

暂无评论

Research on university library service model based on data 2021

Research on university library service model based on data

引用

4th International Conference on Information Systems and computer Aided Education, ICISCAE 2021

作者： Hu, Miningbin Liang, Sha Wuhan Polytechnic College of Computer Technology and Software Engineering Wuhan430070 China

ISBN: (纸本)9781450390255

Information, networking, and digitization have gradually been integrated into all walks of life. New technologies of big data have been born in the era of massive digital information explosion. Big data is used in many aspects of people's lives. In the university library system, the application of data mining and analysis can improve service efficiency and expand development space. Based on data mining technology, this article explores how to use data mining and analysis functions to improve the service model of college libraries and integrate intelligence into the library development path. © 2021 ACM.

关键词： Data mining

来源：评论

学校读者我要写书评

暂无评论

Lessons from the Field: Practical Frameworks for CTF Competition Success

Lessons from the Field: Practical Frameworks for CTF Competi...

引用

Annual IEEE Systems Conference

作者： Dylan Middendorf Suman Bhunia Arthur Carvalho Department of Computer Science and Software Engineering Miami University Oxford Ohio Farmer School of Business Miami University Oxford Ohio

ISBN: (数字)9798331508180

ISBN: (纸本)9798331508197

Capture the Flag (CTF) competitions have become increasingly popular over the previous decade. This research aims to systematically approach organizing CTF competitions by developing two frameworks. The first framework utilizes design science research with a stakeholder approach to generate a definition of success. By defining success, an actionable set of design principles was derived to guide competition organizers. These design principles are structured to provide actionable insight to organizers to help solve the multi-objective problem of stakeholder requirements. The second framework identifies the challenge development life cycle through a temporal approach while describing the current best practices in challenge development. These models were then evaluated through a CTF competition case study hosted for an intra-collegiate audience. This competition utilized both frameworks to assess them in a critical light. To our knowledge, this paper is the first to provide a design science approach for hosting a successful CTF. Based on the findings, a brief discussion on refining the frameworks and potential areas for improvement is provided.

关键词： Adaptation models Systematics Biological system modeling Refining Ecosystems Organizations Stakeholders computer security Best practices Testing

来源：评论

学校读者我要写书评

暂无评论

CTGGAN:A modified generative adversarial network for imbalanced CTG signal classification during labor

CTGGAN:A modified generative adversarial network for imbalan...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Mai, Peilin Feng, Junyuan Li, Luoyi Chen, Qinqun Liu, Guiqing Wei, Hang Guangzhou University of Chinese Medicine School of Medical Information Engineering Guangzhou China Shenzhen University College of Computer Science and Software Engineering Guangzhou China Guangzhou University of Chinese Medicine The First Affiliated Hospital Guangzhou China Guangzhou University of Chinese Medicine Intelligent Chinese Medicine Research Institute Guangzhou China

ISBN: (纸本)9798350386226

Cardiotocography (CTG) is a primary tool for real-time monitoring fetal heart rate (FHR) and uterine contraction (UC) signals during labor. Nowadays CTG signal classification models based on deep learning have facilitated end-to-end CTG interpretation through deep neural networks for automatic feature extraction. However, the extreme rarity of abnormal cases in clinical settings has posed great challenges for CTG deep learning. Meanwhile, most Generative Adversarial Network (GAN)-based studies focused on FHR signal generation and ignored the importance of UC signals during labor. Hence, we introduced CTGGAN, a modified GAN specifically designed to simulate both FHR and UC intrapartum signals simultaneously. Our approach incorporated Wasserstein Distance to prevent gradient vanishing during training, alongside a set of latent code vectors that enhance the generator controllability. We also integrated an auxiliary classifier to impose categorical constraints, thereby enriching the diversity of the generated samples. The experimental results demonstrated that CTGGAN effectively generated CTG signals that closely resembled real CTG signals. In addition, the incorporation of CTGGAN significantly improved performance, with a 19.18% increase in sensitivity over the traditional classification without GAN. Furthermore, the comparison with several state-of-the-art CTG classification models underscored the discriminative performance of our CTGGAN. In conclusion, this study effectively enhanced the detection of abnormal cases for intrapartum fetal monitoring. © 2024 IEEE.

关键词： Neonatal monitoring

来源：评论

学校读者我要写书评

暂无评论

Transforming Payment Processes: A Discussion of AI-Enabled Routing Optimization

Transforming Payment Processes: A Discussion of AI-Enabled R...

引用

Smart Technologies and Systems for Next Generation Computing (ICSTSN), International Conference on

作者： Srinivasa Rao Burri Abhishek Kumar Anupam Baliyan T.Ananth Kumar Senior Software Engineering Manager Western Union Denver CO USA Computer Science and Engineering Chandigarh University Punjab India Computer Science and Engineering IFET College of Engineering Tamilnadu

The utilization of Artificial Intelligence (AI)-enabled software and hardware is expected to enhance the effectiveness of monetary exchanges, which is the main objective of this research. Since transaction costs are based on the actual amount being transferred, customers feel more at ease making purchases using this approach. This work aims to investigate the role of financial systems in effective corporate administration, with a particular emphasis on gateways as key components of these systems. This study investigates the factors (price, procedures, risk assessments, etc.) that influence the connecting of merchant identities across several channels in a digital payment system. When multiple endpoints are connected to a central hub, they are referred to as a ‘‘terminal.’’ The study shows that utilizing AI-enabled software and hardware can reduce transaction costs and increase consumer confidence in purchasing decisions. This research can potentially improve financial systems for businesses and consumers by identifying the most effective applications of AI in this sector.

关键词：

来源：评论

学校读者我要写书评

暂无评论

HiMo: End-to-End Congestion Control for High Speed Rail Data Networking

引用

IEEE Transactions on Intelligent Transportation Systems 2025年第6期26卷 7715-7725页

作者： Chenren Xu Yuhan Zhou Jing Wang Ruihan Li Lingyang Song Guangyu Zhu Key Laboratory of High Confidence Software Technologies Ministry of Education School of Computer Science Peking University Beijing China Zhongguancun Laboratory Beijing China School of Computer Science Peking University Beijing China Department of Electronics Peking University Beijing China School of Automation and Software Engineering Shanxi University Taiyuan Shanxi China

The highly variable nature of cellular networks challenges end-to-end network transmissions in achieving low-latency and high-throughput performance. In high-speed rail (HSR) networks, the intermittent connectivity and capacity dynamics imposed by high client mobility further add complexity and difficulty in providing seamless service. While congestion control algorithms (CCAs) play an essential role in ensuring optimal network performance, prior works on congestion control have predominantly concentrated on enhancing network performance within stationary or low-mobility mobile networks without considering frequent disconnections and highly dynamic network capacities imposed by HSR networks, resulting in severe RTT inflation and slow loss recovery. In this paper, we argue that a dedicated transport layer protocol is necessary for high-mobility scenarios. We propose an end-to-end low-latency congestion control algorithm HiMo for HSR networks that reacts to abrupt bandwidth changes quickly, handles frequent handovers, and is immediately deployable. Our trace-driven emulation on real-world datasets demonstrates that HiMo can reduce 51.3% 95th-percentile latency with comparable throughput on high-speed rail networks, compared to state-of-the-art CCAs.

关键词： Rails Protocols Heuristic algorithms Noise Bandwidth Throughput Market research Rail transportation Low latency communication Intelligent transportation systems

来源：评论

学校读者我要写书评

暂无评论

Imputation with Inter-Series Information from Prototypes for Irregular Sampled Time Series

arXiv

引用

arXiv 2024年

作者： Yu, Zhihao Chu, Xu Ma, Liantao Wang, Yasha Zhu, Wenwu Key Laboratory of High Confidence Software Technologies Ministry of Education Beijing China School of Computer Science Peking University Beijing China Department of Computer Science Tsinghua University Beijing China National Engineering Research Center of Software Engineering Peking University Beijing China

Irregularly sampled time series are ubiquitous, presenting significant challenges for analysis due to missing values. Despite existing methods address imputation, they predominantly focus on leveraging intra-series information, neglecting the potential benefits that inter-series information could provide, such as reducing uncertainty and memorization effect. To bridge this gap, we propose PRIME, a Prototype Recurrent Imputation ModEl, which integrates both intra-series and inter-series information for imputing missing values in irregularly sampled time series. Our framework comprises a prototype memory module for learning inter-series information, a bidirectional gated recurrent unit utilizing prototype information for imputation, and an attentive prototypical refinement module for adjusting imputations. We conducted extensive experiments on three datasets, and the results underscore PRIME’s superiority over the state-of-the-art models by up to 26% relative improvement on mean square error. Our code is available at https://***/yzhHoward/PRIME. Copyright © 2024, The Authors. All rights reserved.

关键词： Mean square error

来源：评论

学校读者我要写书评

暂无评论

Boundary-Guided Learning for Gene Expression Prediction in Spatial Transcriptomics

Boundary-Guided Learning for Gene Expression Prediction in S...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Qu, Mingcheng Wu, Yuncong Di, Donglin Su, Anyang Su, Tonghua Song, Yang Fan, Lei Harbin Institute of Technology Faculty of Computing Harbin China Harbin Institute of Technology School of Astronautics Harbin China Space AI Li Auto Beijing China Jilin University College of Software Changchun China Computer Science and Engineering Unsw Sydney Australia

ISBN: (纸本)9798350386226

Spatial transcriptomics (ST) has emerged as an advanced technology that provides spatial context to gene expression. Recently, deep learning-based methods have shown the capability to predict gene expression from WSI data using ST data. Existing approaches typically extract features from images and the neighboring regions using pretrained models, and then develop methods to fuse this information to generate the final output. However, these methods often fail to account for the cellular structure similarity, cellular density and the interactions within the *** this paper, we propose a framework named BG-TRIPLEX, which leverages boundary information extracted from pathological images as guiding features to enhance gene expression prediction from WSIs. Specifically, our model consists of three branches: the spot, in-context and global branches. In the spot and in-context branches, boundary information, including edge and nuclei characteristics, is extracted using pretrained models. These boundary features guide the learning of cellular morphology and the characteristics of microenvironment through Multi-Head Cross-Attention. Finally, these features are integrated with global features to predict the final *** experiments were conducted on three public ST datasets. The results demonstrate that our BG-TRIPLEX consistently outperforms existing methods in terms of Pearson Correlation Coefficient (PCC). This method highlights the crucial role of boundary features in understanding the complex interactions between WSI and gene expression, offering a promising direction for future research. Codes are available at: https://***/WcloudC0416/BG-TRIPLEX © 2024 IEEE.

关键词： Gene expression

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：