检索结果-内蒙古大学图书馆

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Yunrui Dong Guiduo Duan Tianxi Huang Yunhao Li School of Computer Science and Engineering University of Electronic Science and Technology of China Chengdu China Laboratory of Intelligent Collaborative Computing University of Electronic Science and Technology of China Chengdu China Trusted Cloud Computing and Big Data Key Laboratory of Sichuan Province Chengdu China Chengdu Textile College College of Humanities and General Education Chengdu China School of Information and Software Engineering University of Electronic Science and Technology of China Chengdu China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

In the domain of Multimodal Relation Extraction (MRE), we present the $\color{Red}{\text{W}}$atcher-$\color{Red}{\text{M}}$ediated $\color{Red}{\text{A}}$ttention $\color{Red}{\text{J}}$oint $\color{Red}{\text{L}}$earning Model ($\color{Red}{\text{WMAJL}}$), a novel approach addressing the challenges of modality alignment noise, cross-modal fusion disparity, preservation of textual relative position information, and the distinctiveness of classification labels. WMAJL employs an integrative framework leveraging contrastive learning and variational autoencoder constraints to mitigate modality alignment noise by prioritizing relevant semantic data and effectively reducing extraneous noise that does not contribute to the task. The model’s innovative architecture includes a mediator watcher, which facilitates enhanced cross-modal fusion by enabling nuanced information exchange between textual and visual modalities while preserving the unique characteristics of each modality. Additionally, the design of auxiliary tasks, such as Named Entity Recognition (NER), and output supervision constructs loss functions that preserve relative position information, ensuring a precise depiction of entity relationships throughout the multilayer encoding processes. A key differentiator of WMAJL is its label-centric self-information loss technique, inspired by InfoNCE, which trains the model to cluster similar relation labels in semantically coherent areas, thereby optimizing classification label uniqueness by discerning subtle differences among relation types. The synergistic application of these strategies has led to a significant enhancement of WMAJL’s performance, as evidenced by its state-of-the-art F1 score of $\color{Red}{84.93\%}$ on the MNRE dataset. This achievement surpasses existing benchmarks and sets a new standard for multimodal knowledge extraction, underscoring WMAJL’s potential to revolutionize the MRE landscape.

关键词： Resistance Visualization Technological innovation Noise Autoencoders Semantics Contrastive learning Nonhomogeneous media Speech processing Standards

来源：评论

学校读者我要写书评

暂无评论

Domain-Specific Transport Protocols for In-Network Processing at the Edge: A Case Study of Accelerating Model Synchronization

引用

IEEE Transactions on Mobile computing 2025年

作者： Luo, Shouxi Zhang, Peidong Song, Xin Fan, Pingzhi Xing, Huanlai Luo, Long Yu, Hongfang Southwest Jiaotong University School of Computing and Artificial Intelligence Chengdu611756 China Southwest Jiaotong University Manufacturing Industry Chains Collaboration and Information Support Technology Key Laboratory of Sichuan Province Chengdu611756 China Southwest Jiaotong University Key Laboratory of Information Coding and Transmission Csnmt Int Coop. Res. Centre Chengdu611756 China University of Electronic Science and Technology of China School of Information and Communication Engineering Chengdu611731 China

Nowadays, cross-device federated learning (FL) is the key to achieving personalization services for mobile users and has been widely employed by companies like Google, Microsoft, and Alibaba in production. With the explosive growth in the number of participants, the central FL server, which acts as the manager and aggregator of cross-device model training, would get overloaded, becoming the system bottlenecks. Inspired by the emerging wave of edge computing, an interesting question arises: Could edge clouds help cross-device FL systems overcome the bottleneck? This article provides a cautiously optimistic answer by proposing INP, a FL-specific In-Network Processing framework to achieve the goal. As in-network processing has broken the end-to-end principle of the involved communication and lacks the support of transport protocols, the key is to design domainspecific transport protocols for INP. To fill the gap, we propose the novel Model Download Protocol of MDP and Model Upload Protocol of MUP. With MDP and MUP, edge cloud nodes along the paths in INP can easily eliminate duplicated model downloads and pre-aggregate associated gradient uploads for the central FL server, thus alleviating its bottleneck effect, and further accelerating the entire training progress significantly. © 2025 IEEE.

关键词： Semiconducting indium phosphide

来源：评论

学校读者我要写书评

暂无评论

A Power Control Scheme for Target Localization in WSN

引用

电子学报(英文版) 2008年第3期17卷 443-448页

作者： GU Yu ZHAO Baohua Department of Computer Science University of Science and Technology of China Anhui Province Key Laboratory of Software in Computing and Communication Hefei 230027 China

In this paper, we focus on the target tracking problem in sensor networks and propose an Powersaving target localization scheme (PSTL) based on a conjecture model that reflects the moving pattern of a target, and also a corresponding two-step communication protocol between Base station (BS) and sensors. BS executes a query mechanism to determine which sensors should be used for detailed information according to a limited amount of data received from sensors. This scheme reduces both energy consumption and communication bandwidth requirement, prolongs the lifetime of the wireless sensor networks. Simulation results indicate that it can achieve a high accuracy while saving a large amount of energy.

关键词：无线传感器网络功率控制目标定位推测模型

来源：评论

学校读者我要写书评

暂无评论

LarTap: A Luminance-aware Framework with Text-correlation Priors for Multi-Exposure Image Fusion

引用

IEEE Transactions on Circuits and Systems for Video Technology 2025年

作者： Wang, Enlong Li, Jiawei Yan, Tiantian Lei, Jia Zhou, Shihua Wang, Bin Liu, Jinyuan Kasabov, Nikola K. Dalian University Key Laboratory of Advanced Design and Intelligent Computing Ministry of Education School of Software Engineering Dalian116622 China University of the Science and Technology Beijing School of Computer and Communication Engineering Beijing100083 China Xi’an Jiaotong University School of Software Engineering Shaanxi Xi’an710049 China Dalian Univer sity of Technology School of Mechanical Engineering Dalian116024 China Auckland University of Technology Knowledge Engineering and Discovery Research Institute Auckland1061 New Zealand

Conventional imaging devices often struggle to produce high-dynamic-range (HDR) images that accurately represent natural scenes. To overcome this limitation, multi-exposure image fusion (MEF) techniques have been introduced as a viable solution. Existing MEF approaches aim to enhance performance by optimizing or searching architectures. However, they face challenges in precise feature extraction and scene reconstruction, leading to distortion in the fused images. Additionally, most methods do not adequately address luminance variations across different image regions, which may result in the loss of essential details. To address these challenges, we present a novel luminance-aware MEF framework that integrates text-correlation priors (LarTap). By embedding textual information into fusion process, the proposed framework enhances content extraction and comprehension. Specifically, it consist of two key components: the text-image correlation network (N1) and the multi-exposure fusion network (N2). First, N1 performs correlation training to achieve a holistic alignment between text and image pairs. Its iterative vision encoders (VEs) generate text-correlated prior knowledge to facilitate the fusion process in N2. Second, N2 leverages these priors for scene reconstruction and dynamically adjusts luminance based on comparative perception. Extensive experiments on multiple datasets demonstrate that LarTap outperforms state-of-the-art methods. © 1991-2012 IEEE.

关键词： Image correlation

来源：评论

学校读者我要写书评

暂无评论

Minimum-Energy Broadcast Trees in All-Wireless Networks

引用

电子学报(英文版) 2008年第2期17卷 227-230页

作者： HU Tian ZHAO Baohua Department of Computer Science University of Science and Technology of China Anhui Province Key Laboratory of Software in Computing and Communication Hefei 230027 China

This paper focuses on the problem of constructing the minimum-energy broadcast trees in all-wireless networks. We proposed a greedy algorithm called Greedy maximum-branch replacement algorithm (GMBR) to decrease the total power of the broadcast tree further. This algorithm can be developed to a distributed one easily. Compared with the previous algorithms, GMBR was proved to have better performance, especially when the propagation loss exponent and the network size become larger.

关键词：最小能量广播天线网络技术性能

来源：评论

学校读者我要写书评

暂无评论

NewBalance: Efficient Data Space Management and Algorithmic Optimization for Large-Scale Storage Systems

引用

Chinese Journal of Electronics 2017年第3期26卷 493-501页

作者： XU Guangping LIN Sheng SHI Kai ZHANG Hua Tianjin Key Laboratory of Intelligence Computing and New Software Technology Computer and Communication Engineering School Tianjin University of Technology

Fragmentation usually occurs when data space of original storage nodes has to be reallocated to new added storage nodes during the scale-out evolution of the large-scale storage *** greatly influences its performance and becomes a challenge to manage the whole *** present an efficient space management framework,called New Balance,to reduce fragmentation with the minimum data movement while keeping the storage system load *** space management framework has two phases including the collection phase and the allocation *** the collection phase,we propose a novel algorithm,called the greedy bi-direction collector,which collects enough space for the new storage *** the allocation phase,we formally represent it as a variant of the bin packing problem and then utilize some bin packing heuristics including the first fitting and the best fitting to allocate collected intervals to new added storage *** experimental results show that the amount of intervals can be reduced by 20%;5%and our algorithmic optimization improves the data lookup performance by at least 10%and the scale-out performance by 2X;X.

关键词： Large-scale storage systems Space management Algorithmic optimization Performance evaluation

来源：评论

学校读者我要写书评

暂无评论

Fast Warming Over the Mongolian Plateau a Catalyst for Extreme Rainfall Over North China

引用

Geophysical Research Letters 2025年第8期52卷

作者： Gu, Jun Zhao, Chun Xu, Mingyue Ma, Yuanyuan Hu, Zhiyuan Jin, Chen Guo, Jianping Geng, Tao Cai, Wenju Deep Space Exploration Laboratory/School of Earth and Space Sciences/CMA-USTC Laboratory of Fengyun Remote Sensing/State Key Laboratory of Fire Science/Institute of Advanced Interdisciplinary Research on High-Performance Computing Systems and Software University of Science and Technology of China Hefei China Laoshan Laboratory Qingdao China CAS Center for Excellence in Comparative Planetology University of Science and Technology of China Hefei China Key Laboratory of Land Surface Process and Climate Change in Cold and Arid Regions Northwest Institute of Eco-Environment and Resources Chinese Academy of Sciences Lanzhou China Key Laboratory for Climate Change and Natural Disaster Studies of Guangdong Province School of Atmospheric Sciences Sun Yat-sen University Zhuhai China Southern Marine Science and Engineering Guangdong Laboratory Zhuhai China State Key Laboratory of Severe Weather Chinese Academy of Meteorological Sciences Beijing China and Physical Oceanography Laboratory Ocean University of China Qingdao China

Extreme rainfall events are becoming increasingly severe under a warming climate. North China has experienced several catastrophic rainfall events, of which the rainstorm in 2023 was particularly severe inducing unprecedented damage. Since 1980, the neighboring Mongolian Plateau (MP) has been warming at a rate three times the global average, faster than the surrounding regions. Whether a link exists between extreme rainfall in North China and the fast MP warming is unknown. Here, using global variable-resolution atmospheric model with convection-permitting capability over North China, we find the rapid warming trends, particularly over the MP, are highly conducive to extreme rainfall over North China. In the 2023 case, the fast MP warming induced an anomalous terrestrial high, which in the Western North Pacific Subtropical High created a strong high-pressure system over North China. This system obstructed northeastward movement of Typhoon Doksuri, concentrating moisture supply which prolonged and intensified the extreme. © 2025. The Author(s).

关键词： Global warming

来源：评论

学校读者我要写书评

暂无评论

A Survey of Adversarial Examples in Computer Vision:Attack,Defense,and Beyond

引用

Wuhan University Journal of Natural Sciences 2025年第1期30卷 1-20页

作者： XU keyizhi LU Yajuan WANG Zhongyuan LIANG Chao School of Computer Science Wuhan UniversityWuhan 430072HubeiChina National Engineering Research Center for Multimedia Software(NERCMS) Wuhan UniversityWuhan 430072HubeiChina Key Laboratory of Multimedia and Network Communication Engineering Hubei ProvinceWuhan UniversityWuhan 430072HubeiChina School of Cyber Science and Engineering Wuhan UniversityWuhan 430072HubeiChina

Recent years have witnessed the ever-increasing performance of Deep Neural Networks(DNNs)in computer vision ***,researchers have identified a potential vulnerability:carefully crafted adversarial examples can easily mislead DNNs into incorrect behavior via the injection of imperceptible modification to the input *** this survey,we focus on(1)adversarial attack algorithms to generate adversarial examples,(2)adversarial defense techniques to secure DNNs against adversarial examples,and(3)important problems in the realm of adversarial examples beyond attack and defense,including the theoretical explanations,trade-off issues and benign attacks in adversarial ***,we draw a brief comparison between recently published surveys on adversarial examples,and identify the future directions for the research of adversarial examples,such as the generalization of methods and the understanding of transferability,that might be solutions to the open problems in this field.

关键词： computer vision adversarial examples adversarial attack adversarial defense

来源：评论

学校读者我要写书评

暂无评论

Classifying and clustering in negative databases

引用

Frontiers of Computer Science 2013年第6期7卷 864-874页

作者： Ran LIU Wenjian LUO Lihua YUE School of Computer Science and Technology University of Science and Technology of China Hefei 230027 China Anhui Province Key Laboratory of Software Engineering in Computing and Communication University of Science and Technology of China Hefei 230027 China

Recently, negative databases （NDBs） are proposed for privacy protection. Similar to the traditional databases, some basic operations could be conducted over the NDBs, such as select, intersection, update, delete and so on. However, both classifying and clustering in negative databases have not yet been studied. Therefore, two algorithms, i.e., a k nearest neighbor （kNN） classification algorithm and a k-means clustering algorithm in NDBs, are proposed in this paper, respectively. The core of these two algorithms is a novel method for estimating the Hamming distance between a binary string and an NDB. Experimental results demonstrate that classifying and clustering in NDBs are promising.

关键词： negative databases classification clustering k nearest neighbor k-means hamming distance

来源：评论

学校读者我要写书评

暂无评论

Performance of CO-OFDM system with RS-Turbo concatenated code

引用

Optoelectronics Letters 2015年第6期11卷 457-460页

作者：童峥嵘胡桂彬曹晔张卫华 Key Laboratory of Film Electronics and Communication Devices and Key Laboratory of Intelligent Computing and Novel Software Technology Tianjin University of Technology

In this paper,the RS-Turbo concatenated code is applied to coherent optical orthogonal frequency division multiplexing(CO-OFDM) ***(186,166,8) and Turbo code with code rate of 1/2 are employed for RS-Turbo concatenated *** decoding algorithms,which are Max-Log-MAP algorithm and Log-MAP algorithm,are adopted for Turbo decoding,and the iteration Berlekamp-Massey(BM) algorithm is adopted for RS *** simulation results show that the bit error rate(BER) performance of CO-OFDM system with RS-Turbo concatenated code is significantly improved at high optical signal to noise ratio(OSNR),and the iteration number is reduced compared with that of the Turbo coded ***,when the Max-Log-MAP algorithm is adopted for Turbo decoding,the transmission distance of CO-OFDM system with RS-Turbo concatenated code can reach about 400 km without error,while that of the Turbo coded system can only reach about 240 km when BER is lower than 10^(-4) order of magnitude.

关键词： OFDM系统 RS解码 Max-Log-MAP算法性能 Turbo码正交频分复用解码算法光信噪比

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：