检索结果-内蒙古大学图书馆

13th International Conference on Computing and Pattern Recognition, ICCPR 2024

作者： Li, Zongshu Zhu, Guibo Yi, Dongyi Wang, Jinqiao Foundation Model Research Center Institute of Automation Chinese Academy of Sciences Beijing China Algorithm Development Department Wuhan Artificial Intelligence Research Wuhan Hubei China

ISBN: (纸本)9798400717482

Mamba - the selective SSM (S6) model, makes up for inherent shortcomings of Transformer in long sequence processing and has been gaining significant attention from researchers. In this paper, we employ Mamba to build a multimodal chat model - Multimodal Mamba. Inspired by the prevalent structure of multimodal models, MM-Mamba consists of three parts: vision encoder, multimodal projector and language decoder. Firstly, we simply replace the text decoder from Transformer to Mamba. Secondly, considering that parameters and pretraining tokens of Mamba are significantly less than Transformer, we propose a multi-stage training pipeline to train MM-Mamba. Finally, in order to further boost the proficiency of MM-Mamba across a wide range of multimodal downstream tasks, we introduce several powerful techniques, such as higher input image resolution, noisy embedding finetuning. Compared with multimodal large language models (MLLMs) built upon Transformer, MM-Mamba can be adapted to various downstream tasks with extremely low temporal and economic training cost. Experimental results showed that albeit with notably fewer parameters and significantly limited training data volume, remarkable results are achieved by MM-Mamba not only on English benchmarks but also in Chinese downstream tasks. Code is released at here. © 2024 Copyright held by the owner/author(s).

关键词： Mamba

来源：评论

学校读者我要写书评

暂无评论

Decomposition and comparative analysis of urban-rural differences in eHealth literacy among nursing students: A multisite cross-sectional study

引用

Nurse Education Today 2025年 146卷 106562页

作者： Wang, Xue Ge, Lina Yu, Zhenjie Qiao, Qiao Guo, Xiaoming Liu, Shiping Zhang, Chengcai Zang, Shuang Department of Community Nursing School of Nursing China Medical University Shenyang China Department of Obstetrics and Gynecology Shengjing Hospital of China Medical University Shenyang China School of Nursing Tianjin Medical University Tianjin China Department of Radiation Oncology The First Hospital of China Medical University Shenyang China Algorithm Research and Development Department Neusoft IntelliRay Technology Co. Ltd. Shenyang China

Background: Investigating the urban-rural differences in eHealth literacy among Chinese nursing students is imperative for enhancing healthcare education and practice. This study aimed to investigate the factors that contribute to urban-rural differences in eHealth literacy among Chinese nursing students and measure their respective contributions. Design: A multisite cross-sectional study. Settings: In China. Participants: A total of 967 nursing students were involved in this study. Methods: Multivariate linear regression models were used to explore the associations between sociodemographic and health-related factors and the eHealth literacy of Chinese nursing students. Additionally, Blinder-Oaxaca decomposition was employed to scrutinize the factors attributing to the urban-rural differences in eHealth literacy among Chinese nursing students, elucidating their respective contributions. Results: The mean eHealth literacy score for urban nursing students (mean: 20.26, SD: 3.61) exceeded that of rural nursing students (mean: 19.53, SD: 3.56) (p < 0.001). Blinder-Oaxaca decomposition analysis indicated that 84.82 % of the urban-rural differences in eHealth literacy among nursing students can be explained. The urban-rural differences are primarily explained by family health (40.05 %). Conclusions: The study emphasizes the importance of tailoring educational curricula to address urban-rural differences in eHealth literacy among nursing students in China. Specifically, incorporating content related to family health in educational programs is crucial. © 2024 Elsevier Ltd

关键词： China Comparative analysis Decomposition analysis eHealth literacy Nursing students Urban-rural difference

来源：评论

学校读者我要写书评

暂无评论

DM Code Key Point Detection algorithm Based on CenterNet

引用

Computers, Materials & Continua 2023年第11期77卷 1911-1928页

作者： Wei Wang Xinyao Tang Kai Zhou Chunhui Zhao Changfa Liu School of Information Engineering Chang’an UniversityXi’an710018China Algorithm Research and Development Department GRGBanking Equipment Co.Ltd.Guangzhou510663China Research and Development Department Xi’an Soar Electromechanical TechnologyLtd.Xi’an710043China

Data Matrix(DM)codes have been widely used in industrial *** reading of DM code usually includes positioning and *** positioning is a prerequisite for successful *** image processing methods have poor adaptability to pollution and complex *** deep learning-based methods can automatically extract features,the bounding boxes cannot entirely fit the contour of the *** image processing methods are required for precise positioning,which will reduce *** of the above problems,a CenterNet-based DM code key point detection network is proposed,which can directly obtain the four key points of the DM *** with the existing methods,the degree of fitness is higher,which is conducive to direct *** further improve the positioning accuracy,an enhanced loss function is designed,including DM code key point heatmap loss,standard DM code projection loss,and polygon Intersection-over-Union(IoU)loss,which is beneficial for the network to learn the spatial geometric characteristics of DM *** experiment is carried out on the self-made DM code key point detection dataset,including pollution,complex background,small objects,etc.,which uses the Average Precision(AP)of the common object detection metric as the evaluation *** reaches 95.80%,and Frames Per Second(FPS)gets 88.12 on the test set of the proposed dataset,which can achieve real-time performance in practical applications.

关键词： DM code key point detection CenterNet object detection enhanced loss function

来源：评论

学校读者我要写书评

暂无评论

Deterministic and Stochastic Simulators for Non-Isotropic V2V-MIMO Wideband Channels

引用

China Communications 2018年第7期15卷 18-29页

作者： Yiran Li Xiang Cheng Nan Zhang State Key Laboratory of Advanced Optical Communication Systems and Networks School of Electronics Engineering and Computer Science Peking University Wireless Algorithm Department Product Research and Development System ZTE Corporation

In this paper, we consider a novel two-dimensional(2D) geometry-based stochastic model(GBSM) for multiple-input multiple-output(MIMO) vehicle-to-vehicle(V2V) wideband fading channels. The proposed model employs the combination of a two-ring model and a multiple confocal ellipses model, where the signal is sum of the line-of-sight(Lo S) component, single-bounced(SB) rays, and double-bounced(DB) rays. Based on the reference model, we derive some expressions of channel statistical properties, including space-time correlation function(STCF), Doppler spectral power density(DPSD), envelope level crossing rate(LCR) and average fade duration(AFD). In addition, corresponding deterministic and stochastic simulation models are developed based on the reference model. Moreover, we compare the statistical properties of the reference model and the two simulation models in different scenarios and investigate the impact of different vehicular traffic densities(VTDs) on the channel statistical properties of the proposed model. Finally, the great agreement between simulation models and the reference model demonstrates not only the utility of simulation models, but also the correctness of theoretical derivations and simulations.

关键词： vehicle-to-vehicle wideband channels simulation model statistical properties

来源：评论

学校读者我要写书评

暂无评论

IGP Convergence Optimization with AI 24

IGP Convergence Optimization with AI

引用

24th IEEE International Conference on Communication Technology, ICCT 2024

作者： Das, Biswarup Prihodko, Maxim Krantsevich, Artem Rakhimov, Abdurasul Jin, Jian Yang, Pingan Dai, Longfei Huawei Moscow Network Algorithm Laboratory Moscow Russia Huawei Data Communication Protocol Development Department Beijing China

ISBN: (纸本)9798350363760

This paper develops an Interior Gateway Protocol (IGP) convergence optimization strategy and proposes an AI-based method to predict a temporary Routing Information Base (RIB) table with high accuracy to be used for packet forwarding while the exact RIB is being computed using a traditional approach, for example Dijkstra algorithm. The proposed method consists of AI-model and the training procedure and considers RIB prediction as a classification task when each Destination in a network is assigned to a class representing corresponding Next Hop. In this work, the complete IGP convergence optimization strategy is implemented and tested in a physical environment for 2 different IGP protocols, namely OSPF and IS-IS. As in traditional IGP process, the AI model takes Link State DataBase (LSDB) and incoming Link State Advertisement (LSA) messages as the input and synthesizes a temporary RIB in the required internal format which is used then for packet forwarding until traditional IGP process finishes exact RIB recomputation. The proposed method is implemented and tested in a physical testbed environment emulating the networks with up to 10000 network elements (routers). The performance evaluation shows methods' ability to predict RIB table with high accuracy and speed, leading to faster IGP convergence which is observed as 2 to 5 times faster packet-loss elimination in the network after a link failure accident. © 2024 IEEE.

关键词： Gateways (computer networks)

来源：评论

学校读者我要写书评

暂无评论

Extended kalman filter based robust altitude controller for sea skimming missiles

Extended kalman filter based robust altitude controller for ...

引用

AIAA Guidance, Navigation, and Control Conference, 2016

作者： Dülgar, Özgün Gezer, Rüştü Berk Kutay, Ali Türker Algorithm Development Department Roketsan Inc. Ankara06780 Turkey Department of Aerospace Engineering Middle East Technical University Ankara06800 Turkey

ISBN: (数字)9781624103896

ISBN: (纸本)9781624103896

A sea skimming missile is needed to be flown above mean sea level as low as it can, in order to decrease detectability;so that, survivability of the missile against counter attacks of the target ships is maximized. On the other hand, flying at very low height is a tough task under disturbances due to sea waves and measurement errors of various sensors used in height control loop. Thus, a robust altitude control system design is the main challenge among the other control algorithms for a sea skimming missile. In this paper, extended Kalman filter based altitude controller method is proposed and compared with the existing designs in literature. Simulation results under a few scenarios, in which sensor errors, severe sea conditions, and limited computing power are taken into account, are shared. Results show that the extended Kalman filter based altitude controller has satisfactory performance against many real world issues. © 2016, American Institute of Aeronautics and Astronautics Inc, AIAA. All rights reserved.

关键词： Extended Kalman filters

来源：评论

学校读者我要写书评

暂无评论

Reinforcement Learning Based End-to-end Autonomous Obstacle Avoidance for Manipulators

Reinforcement Learning Based End-to-end Autonomous Obstacle ...

引用

2023 China Automation Congress, CAC 2023

作者： Zhang, WenPeng Chen, Dan Sun, Xiao Huang, WeiBing Xu, ZheZhuang Fuzhou University School of Electrical Engineering and Automation Fuzhou China Shenzhen Evomotion Co. Ltd Algorithm Development Department Shenzhen China

ISBN: (纸本)9798350303759

In this paper, a simple end-to-end obstacle avoidance method is introded for manipulators. First of all, the 2D images in the workspace are used to uniformly describe the characteristics of obstacles in the 3D space. After that, a model-free reinforcement learning algorithm DrQ-v2 is used to train the obstacle avoidance strategy, which directly outputs the joint angles to avoid the obstacles autonomously and accurately in the joint space. Finally, the simulation results demonstrate that the proposed end-to-end simple control method is more effective and convenient in handling obstacle avoidance tasks in complex dynamic environments. © 2023 IEEE.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Field Trial of a New Metropolitan WDM Architecture and Flexible Scheduling based on Optical Layer OAM

Field Trial of a New Metropolitan WDM Architecture and Flexi...

引用

2024 Asia Communications and Photonics Conference, ACP 2024 and International Conference on Information Photonics and Optical Communications, IPOC 2024

作者： Zhao, Yang Liu, Yucong Li, Xiang Wang, Dong Hu, Changbin Zhang, Dechao Department of Fundamental Network Technology China Mobile Research Institute Beijing China Optical System & Algorithm Development Department Huawei Technologies Co. Ltd. Wuhan China

A novel wavelength pooling technology and a flexible scheduling method based on optical layer OAM (O-OAM) are proposed. Costs of network construction and maintenance are reduced significantly, and service latency is r... 详细信息

ISBN: (纸本)9798350379266

关键词： Centralized Control Metropolitan Transport Network Optical Layer OAM Wavelength Pooling

来源：评论

学校读者我要写书评

暂无评论

Field Trial of End-to-end Management and Control of Semiactive WDM System for 5G Centralized Front-haul Network

Field Trial of End-to-end Management and Control of Semiacti...

引用

2023 Optical Fiber Communications Conference and Exhibition, OFC 2023

作者： Zhao, Yang Sun, Jiang Zheng, Haomian Yu, Chaode Wang, Dong Cai, Qian Li, Yunbo Han, Liuyan Zhang, Dechao Li, Han China Mobile Research Institute Department of Network and It Technology Beijing100053 China Huawei Technologies Co. Ltd Optical System & Algorithm Development Department Wuhan430223 China

A novel management and control architecture and abstract model of semi-active WDM system for 5G front-haul network are proposed. The functions of device information acquisition, alarm monitoring and fault location is ... 详细信息

ISBN: (纸本)9781957171180

关键词： 5G mobile communication systems

来源：评论

学校读者我要写书评

暂无评论

Fast Rail Surface Inspection Method and System Based on 3D Line Laser Imaging 30

Fast Rail Surface Inspection Method and System Based on 3D L...

引用

30th International Conference on Mechatronics and Machine Vision in Practice, M2VIP 2024

作者： Liu, Lingshu Kang, Yihua Liu, Fuqiang Liu, Shiyuan Huazhong University of Science and Technology School of Mechanical Science and Engineering Wuhan China Software Algorithm Department Beijing Metallurgical Technology Development Co. LTD. Beijing China

ISBN: (纸本)9798350391916

Due to the significance and complexity of rail surface quality inspections on production lines, manual completion is inefficient. Existing 3D vision techniques, such as stereo vision, are too complex and cumbersome for industrial use. This paper introduces a straightforward and fast line laser calibration method, along with a rail surface defect detection approach using line laser 3D reconstruction. By projecting line lasers onto the sides of a calibration board of specific thickness, the position of the laser centerlines in the side areas is extracted to ascertain the world coordinates of the corresponding point clouds. Four sets of line laser measurement devices are then employed to conduct a full coverage scan of the rail. A streamlined multi-camera joint calibration method is presented, which stitches together 3D point clouds from various perspectives to recreate the complete cross-sectional profile of the rail. After transforming into a surface depth variation map, defects are identified based on depth thresholds. The method not only precisely locates and categorizes defects but also reduces reconstruction errors by 0.1mm and simplifies the calibration process by eliminating the need for multiple rotations of calibration targets, thereby speeding up and enhancing the feasibility of industrial application. The approach overcomes the complexities and cumbersome processes associated with traditional methods not suited for industrial settings. The system can detect at a speed of 1.5m/s, identifying surface defects deeper than 0.3mm with a verified accuracy of less than 0.02mm, thus fulfilling the requirements for rail surface defect inspection. © 2024 IEEE.

关键词： Inspection

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：