检索结果-南通市图书馆

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

基于相似度加速的自适应聚类联邦学习

通信学报 2024年第3期45卷 197-207页

作者：朱素霞顾玢珂孙广路哈尔滨理工大学计算机科学与技术学院黑龙江哈尔滨150080 黑龙江省智能信息处理及应用重点实验室黑龙江哈尔滨150080

为了解决联邦学习过程中数据异质性导致模型性能下降的问题,考虑对联邦模型个性化,提出了一种新的基于相似度加速的自适应聚类联邦学习(ACFL)算法,基于客户端本地更新的几何特性和客户端联邦时的正向反馈实现自适应加速聚类,将客户端划... 详细信息

为了解决联邦学习过程中数据异质性导致模型性能下降的问题,考虑对联邦模型个性化,提出了一种新的基于相似度加速的自适应聚类联邦学习(ACFL)算法,基于客户端本地更新的几何特性和客户端联邦时的正向反馈实现自适应加速聚类,将客户端划分到不同任务簇,同簇中数据分布相似的客户端协同实现聚类联邦学习(CFL),从而提升模型性能。该算法不需要先验确定类簇数量和迭代划分客户端,在避免现有基于聚类的联邦算法计算成本过高、收敛速度慢等问题的同时保证了模型性能。在常用数据集上使用深度卷积神经网络验证了ACFL的有效性。结果表明,所提算法性能与聚类联邦学习算法相当,优于传统的迭代联邦聚类算法(IFCA),且具有更快的收敛速度。

关键词：联邦学习个性化聚类几何特性正向反馈

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

电力系统融入新能源的运行与控制

今日自动化 2024年第2期 61-63页

作者：孙光禄张伟孙磊北京京能国际控股有限公司北方分公司内蒙古呼和浩特010051

由于新能源具有间歇性和波动性,其发电具有不稳定性和随机性,对电力系统的安全稳定运行提出了挑战。文章针对电力系统融入新能源的运行与控制方式展开详细分析,为全面促进电力系统稳定运行提供参考。

关键词：电力系统新能源运行控制

Spam Filtering:Online Naive Bayes Based on TONE

维普期刊数据库博看期刊评论

在线全文

学校读者我要写书评

暂无评论

ZTE Communications 2013年第2期11卷 51-54页

作者： guanglu sun Hongyue sun Yingcai Ma Yuewu Shen Research Institute of Information Technology Tsinghua University ZTE Corporation School of Computer Science and Technology Harbin University of Science and Technology

The naive, Bayes （NB） model has been successfully used to tackle spare, and is very accurate. However, there is still room for improwment. We use a train on or near error （TONE） method in online NB to enhance the perfornmnee of NB and reduce the number of training emails. We conducted an experiment to determine the performanee of the improved algorithm by plotting （I-ROCA）% curves. The resuhs show that the proposed method improves the performanee of original NB.

关键词： spare fihering online naive Bayes train-on or near error

基于质量扰动鹈鹕优化算法的图像匹配方法研究

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

郑州大学学报（理学版） 2024年第4期56卷 81-87页

作者：杨光露胡宏帅王小明冯绍志王凤仙孙俊峰郑州轻工业大学电气信息工程学院河南郑州450002 河南中烟工业有限责任公司河南郑州450016 广西中烟工业有限责任公司广西南宁530000

针对传统群优化算法在图像匹配中存在调节参数多、不易操作等问题,提出了一种基于质量扰动的鹈鹕优化算法(disturbance quality pelican optimization algorithm,DPOA)的图像匹配方法。传统的鹈鹕优化算法(pelican optimization algorit... 详细信息

针对传统群优化算法在图像匹配中存在调节参数多、不易操作等问题,提出了一种基于质量扰动的鹈鹕优化算法(disturbance quality pelican optimization algorithm,DPOA)的图像匹配方法。传统的鹈鹕优化算法(pelican optimization algorithm,POA)在求解多峰函数问题时,其全局收敛性需要进一步改进。首先,引入了一种新的质量扰动方法,通过检测分布点附近的点来收敛到更好的解,提高了在解决多峰函数问题时易陷入局部最优的问题,同时提高了算法的收敛精度。其次,通过数据集CEC2019对算法的有效性进行评价。最后,通过提取图像的方向梯度直方图(histogram of oriented gradient,HOG)将DPOA算法在图像匹配中应用,并通过实验仿真,证明了DPOA算法在图像匹配中的可行性与有效性。

关键词：图像匹配鹈鹕优化算法质量扰动方向梯度直方图群优化

Computer English Acquisition Environment Construction Based on Question Answering Technology Based on Question Answering Technology

同方期刊数据库博看期刊评论

在线全文

学校读者我要写书评

暂无评论

国际计算机前沿大会会议论文集 2016年第2期 3-5页

作者： Fei Lang Peipei Li Jian Kang guanglu sun School of Foreign Languages Harbin University of Science and TechnologyHarbinHeilongjiangChina Beijing Institute of Astronautical Systems Engineering BeijingChina School of Computer Science and Technology Harbin University of Science and TechnologyHarbinHeilongjiangChina

In-depth vocabulary knowledge plays a key role in L2(second language)vocabulary acquisition,and computer English vocabulary is vital to academic performances of students majored in Computer Science.While many Chinese students cannot realize computer English vocabulary acquisition due to lacking input environment of computer English.To solve this problem,with techniques of keyword search,information retrieval,and text classification,to meet the diverse learning needs of students,it built the in-depth computer English vocabulary knowledge acquisition environment for the high frequency of computer English words,providing the depth of knowledge such as polysemy,synonyms,syntax,and pragmatics.The construction of in-depth computer English vocabulary acquisition environment helps future studies on computer English teaching applications.

关键词： Computer English In-depth vocabulary knowledge Information retrieval Text classification

Training Mode of Personnel Majoring in Network Engineering Based on Three Main Lines

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

国际计算机前沿大会会议论文集 2016年第2期 33-34页

作者： Hongzhuo Qi guanglu sun Zhiyong Luo School of Computer Science and Technology Harbin University of Science and TechnologyHarbin 150080HeilongjiangChina

Network engineering is an emerging major across many disciplines,so there is yet to be an established personnel training program for this field.Moreover,there is a lack of reference material concerning personnel training.Therefore,it is especially urgent to explore a model suitable for personnel training in network engineering and study relevant personnel training systems.In this paper,the present situation of network engineering personnel training in China is analyzed,and basic orientation and training targets of network engineering major are explored.We find that network engineering personnel training is based on three main combinations:theory and practice,learning and innovation under the social computing environment,and classroom and employment experience,thereby providing constructive experience for training high-quality personnel specializing in network engineering.

关键词： Network engineering Personnel training Social computing Teaching practice

Traffic Collection and Analysis System

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

国际计算机前沿大会会议论文集 2016年第2期 67-69页

作者： Jinlai Liu Haitao Wen Xiangyu Hou guanglu sun Suxia Zhu School of Computer Science and Technology Research Center of Information Security and Intelligent TechnologyHarbin University of Science and TechnologyHarbinChina

This paper designs and implements a network traffic collection and analysis system(NTCA).Network traffic is collected from multiple clients.According to the offline traffic analysis,network flows are analyzed and annotated with application protocol.At the same time,the results are stored in the system.NTCA also provides a visual interface to simplify the configuration process.With good flexibility and scalability,NTCA can be extended to a network traffic clas sification system,which is crucial to the research of network security technology.

关键词： Traffic classification Traffic collection and analysis Feature extraction Traffic annotation

Computer English Teaching Based on WeChat

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

国际计算机前沿大会会议论文集 2016年第2期 6-7页

作者： Fei Lang Kexin Zhang Peipei Li guanglu sun School of Foreign Languages Harbin University of Science and TechnologyHarbinHeilongjiangChina School of Computer Science and Technology Harbin University of Science and TechnologyHarbinHeilongjiangChina

Computer English plays a vital role in academic study for students majored in Computer Science.As the fast development of information technology,traditional English teaching methods cannot meet students’learning needs.WeChat is the most popular mobile application of social networks in China.Its features of real-time communication,interactivity and utilization of fragmentary time attract educators’attention to its application in teaching.This paper carried out an empirical study to explore applying WeChat in Computer English teaching.With WeChat,an English interaction environment was built,and interactive learning activities were implemented.The experimental results show teaching based on WeChat is effective to improve students’Computer English skills.

关键词： Computer English WeChat Social networks English teaching

A Novel Approach for the Identification of Morphological Features from Low Quality Images

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

国际计算机前沿大会会议论文集 2016年第1期 12-14页

作者： Weiqiang Xia Zhijun Hu Huijuan Zhai Jian Kang Jingqun Song guanglu sun Beijing Institute of Astronautical Systems Engineering BeijingChina School of Computer Science and Technology Harbin University of Science and TechnologyHarbinHeilongjiangChina

In this paper, a novel mathematical morphological approach is proposed, which is combined with an active threshold-based method for the identification of morphological features from images with poor qualities. The algorithm is very fast and needs low computing power.First, a mixed smooth filtering is designed to remove background noises. Second, an active threshold-based method is discussed to create a binary image to achieve rough segmentation. Third, some simple morphological operations, such as opening, closing, filling, and so on, are designed and applied to get the final result of segmentation. After morphological analysis, morphological features, such as contours, areas, numbers, locations, and so on, are obtained. Finally, the comparisons with other conventional methods validate the effectiveness, and an additional experimental result proves the repeatability of the proposed method.

关键词： Mathematical morphology Automatic image segmentation and analysis Active threshold Morphological operator