检索结果-内蒙古大学图书馆

IEEE International Conference on Multimedia and Expo (ICME)

作者： Jiquan Peng Chaozhuo Li Yi Zhao Yuting Lin Xiaohan Fang Jibing Gong School of Information Science and Engineering Yanshan University Qinhuangdao China The Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Microsoft Research Asia Beijing China

Vision transformers have significantly advanced the field of computer vision in recent years. The cornerstone of these transformers is the multi-head attention mechanism, which models interactions between visual elements within a feature map. However, the vanilla multi-head attention paradigm independently learns parameters for each head, which ignores crucial interactions across different attention heads and may result in redundancy and under-utilization of the model’s capacity. To enhance model expressiveness, we propose a novel nested attention mechanism, Ne-Att, that explicitly models cross-head interactions via a hierarchical variational distribution. We conducted extensive experiments on image classification, and the results demonstrate the superiority of Ne-Att.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A hierarchical residual network with compact triplet-center loss for sketch recognition

arXiv

引用

arXiv 2021年

作者： Wang, Lei Zhang, Shihui He, Huan Zhang, Xiaoxiao Sang, Yu School of Information Science and Engineering Yanshan University Hebei Qinhuangdao066004 China Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Hebei Qinhuangdao066004 China School of Mathematics and Information Science and Technology Hebei Normal University of Science and Technology Qinhuangdao066004 China

With the widespread use of touch-screen devices, it is more and more convenient for people to draw sketches on screen. This results in the demand for automatically understanding the sketches. Thus, the sketch recognition task becomes more significant than before. To accomplish this task, it is necessary to solve the critical issue of improving the distinction of the sketch features. To this end, we have made efforts in three aspects. First, a novel multi-scale residual block is designed. Compared with the conventional basic residual block, it can better perceive multi-scale information and reduce the number of parameters during training. Second, a hierarchical residual structure is built by stacking multi-scale residual blocks in a specific way. In contrast with the single-level residual structure, the learned features from this structure are more sufficient. Last but not least, the compact triplet-center loss is proposed specifically for the sketch recognition task. It can solve the problem that the triplet-center loss does not fully consider too large intra-class space and too small inter-class space in sketch field. By studying the above modules, a hierarchical residual network as a whole is proposed for sketch recognition and evaluated on Tu-Berlin benchmark thoroughly. The experimental results show that the proposed network outperforms most of baseline methods and it is excellent among non-sequential models at present. Copyright © 2021, The Authors. All rights reserved.

关键词： Touch screens

来源：评论

学校读者我要写书评

暂无评论

An improved algorithm for detecting community structure based on node similarity

引用

Journal of Computational Information systems 2014年第9期10卷 3805-3813页

作者： Guo, Jingfeng Wei, Aiying Lv, Jiaguo Liu, Zhen College of Information Science and Engineering Yanshan University Qinhuangdao 066004 China Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Qinhuangdao 066004 China College of Information Science and Engineering Zaozhuang University Zaozhuang 277100 China

Detecting community structure based on node similarity cost lower time complexity, but they ignore the indirect relationships of nodes. We proposed an improved algorithm for detecting community structure based on node similarity (AINS). The algorithm presents a neighbor subgraph model based on akincore node pair, considering the direct and indirect relations between nodes. It introduced node indirect similarity to solve the problem which only pay attention to direct relationships, ensuring that the results meet the common definition of community structure (the number of edges between community is greater than that within community). The experimental results show that the AINS algorithm can obtain better community classification result effectively. © 2014 Binary Information Press.

关键词： Information systems

来源：评论

学校读者我要写书评

暂无评论

Frequent closed itemsets mining using ITBitree

引用

International Journal of Advancements in Computing technology 2012年第17期4卷 271-279页

作者： Ren, Jiadong Song, Wei Yu, Shiying College of Information Science and Engineering Yanshan University Qinhuangdao Hebei 066004 China The Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Qinhuangdao City 066004 China Hebei Provincial S and T Management Information Center 159 Dongfeng Road Shijiazhuang Hebei 050021 China

The set of frequent closed itemsets is much smaller than the one of all frequent itemsets. Meanwhile it retains the completeness. In this paper, we propose an ITBitree-based algorithm, called ITBitreeFCIM (ITBitree Frequent Closed Itemsets Miner), for mining the frequent closed itemsets directly. A novel structure termed ITBitree (Itemset Tidset Binary Tree) is defined to store the transaction and itemset information. Frequent closed itemsets are generated directly by searching the right nodes in each layer and the right nodes in their respective search space deeply from this tree using top-down strategy. For the items with same support in the adjacent layers or the items with the same Tids, we use item merging technique to prune the search space. The searched space is used to avoid the generation of duplicates and enormous closure checks. In the process we do not need to check whether a candidate itemset is closed, so we save a lot of time. The experiment proofs the efficiency of the proposed approach.

关键词： Data mining

来源：评论

学校读者我要写书评

暂无评论

K-means sequence clustering algorithm based on Two-stage center sequences selection

引用

Journal of Computational Information systems 2014年第6期10卷 2351-2358页

作者： Wu, Di Ren, Jiadong College of Information Science and Engineering Yanshan University Qinhuangdao 066004 China Department of Information and Electronic Engineering Hebei University of Engineering Handan 056038 China Key Lab of Computer Virtual Manufacturing and System Integration of Hebei Province Yanshan University Qinhuangdao 066004 China

K-means sequence clustering is a hot topic in data mining. However, the clustering quality of K-means is affected heavily by outliers and noises. In this paper, a new sequence similarity measurement, sequence density and Q neighborhoods of sequence are deffned and then a novel K-means sequence clustering algorithm TCSSK-means (Two-stage center sequences selection-based K-means) is proposed. In the first stage, the sequence with the minimum density neighborhood is viewed as initial clustering center, and the distances between each initial clustering center are as far as possible, the KICCSS (K initial clustering center sequences set) is obtained. In the second stage, by selecting the mean sequence of the larger similarity sequences set, which meets certain conditions with clustering centers in the T-1-th round, K clustering centers of sequence databases in the T-th round are updated. Repeat the renewal process of sequence clustering centers until the sequence clustering criterion function converges. K final sequence clusters are gained finally. The experimental results and analysis show that the performance of TCSSK-means is better in clustering quality and running time. Copyright © 2014 Binary Information Press.

关键词： Data mining

来源：评论

学校读者我要写书评

暂无评论

A new estimator analysis method to synchronization among multivariate neural series of mild cognitive impairment

International Journal of Digital Content Technology and its ...

引用

International Journal of Digital Content technology and its Applications 2012年第23期6卷 868-874页

作者： Wen, Dong Guan, Xinyong Chen, Ruihong The Key Laboratory for Computer Virtual Technology System Integration of HeBei Province Yanshan University Qinhuangdao Hebei 066004 China Liren College Yanshan University Qinhuangdao Hebei 066004 China Division of General Internal Medicine Xiantao First People's Hospital Xiantao Hubei 433000 China

Synchronization decay of resting state EEG has shown that cognitive dysfunction in mild cognitive impairment (MCI) was relevant to a loss of functional connectivity in intermediate frequency bands. A new estimator analysis method called new S estimator (NSE) proposed recently by us quantifies synchronization between neuronal signals at multiple *** paper meant to explore the behavior of synchronization of multichannel EEG in MCI and healthy normal controls at rest, and preliminarily make clear the clinical significance of the NSE in MCI *** (EEG) were recorded from 10 MCI patients and 12 age-matched healthy normal controls (NC). NSE values were computed both across the all frequency band and separately in the delta, theta, alpha, beta (including beta1, beta2 and beta3), and gamma bands. The Montreal Cognitive Assessment (MoCA) was used to assess the symptom severity of MCI patients and *** values in the alpha and beta1 bands were significantly lower in MCI patients than in NC. NSE values in the alpha and beta1 bands were positively correlated with the MoCA scores in all participants (MCI and NC). In MCI patients, NSE values in the alpha and beta1 bands were also positively correlated with MoCA *** results suggest that NSE values are a useful correlate of EEG synchronization in MCI patients.

关键词： Synchronization

来源：评论

学校读者我要写书评

暂无评论

An algorithm based on bit vector decomposition and hash linklist for mining frequent patterns in data streams

International Journal of Digital Content Technology and its ...

引用

International Journal of Digital Content technology and its Applications 2012年第22期6卷 570-579页

作者： Ren, Jiadong Zhang, Yuanyuan Yu, Shiying College of Information Science and Engineering Yanshan University Qinhuangdao Hebei 066004 China Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Qinhuangdao City 066004 China Hebei Provincial S and T Management Information Center 159 Dongfeng Road Shijiazhuang 050021 China

Most of algorithms based on bit vector for mining frequent pattern always produce candidate itemsets and scan the same data repeatedly;this increases the running time and the space consumption. In this paper, an algorithm BVDAHL based on bit vector decomposition and hash linklist for mining frequent patterns on data streams which solves the above issues, is proposed. The itemsets (transactions) whose corresponding number of 1 is k, denoted as k-one itemsets (k-one transactions). The arrival transactions are converted into bit vectors, and permutations and combinations are used to decompose converted results, then the decomposed itemsets are stored in the hash linklist. In hash function, according to the position of 1 of bit vector to compute the address, if the transaction has same address and same content that already exists in data domain, then the value adds one in count domain, otherwise the pointer in pointer domain points to the next node directly, it is no need to deal with conflict. Afterwards the property of anti-monotonic will be used to prune the k+1-one transactions in bit vector table if there exists infrequent k-one itemsets that decomposed by such k+1-one transactions. Then the algorithm repeats the above processes until the transaction in bit vector table can not be decomposed. Finally we will receive the frequent itemsets by comparing the obtained value of count domain in hash linklist with minSup. Experiment results show that BVDAHL is very efficiency and scalab.e.

关键词： Hash functions

来源：评论

学校读者我要写书评

暂无评论

Adaptive multilevel meshes for cloth simulation

引用

Journal of Information and Computational Science 2014年第11期11卷 3865-3872页

作者： Zhao, Jing Pei, Shanshan Lv, Mengya Li, Ying Tang, Yong Yang, Meiling Zhang, Mingmin Pan, Zhigeng The College of Information Science and Engineering Yanshan University Qinhuangdao China The Key Laboratory for Computer Virtual Technology and System Integration of HeBei Province Qinhuangdao China The College of Computer Science and Technology Zhejiang University Hangzhou China Institute of Service Engineering Hangzhou Normal University Hangzhou China

A novel method based on multilevel meshes is proposed to simulate the cloth draping process rapidly on a complex model. Considering of the difficulty of computing the realtime collision between cloth and complex models by normally methods, the draping process of cloth is divided into several phases and at each phase the mesh size is smaller than its previous one. The optimized 1-to-4 split is used to drive the rapid changes in the grid hierarchy and to simplify the calculations. We choose the widely used ellipsoid as the bounding cell to enclose the complex models in the environment. The ellipsoid avoids the complex structure of hierarchical bounding box and eliminates the impossible collision model area rapidly. Experiments demonstrate that this method can accelerate the draping process of cloth and decrease the number of iterations and the calculation of collision detection. Further mesh refinement improves the reality of cloth when it reaches the final position and ensures the realism of the simulation in real time. ©2014 Binary Information Press

关键词： Mesh generation

来源：评论

学校读者我要写书评

暂无评论

A Practical Cloud API Complementary Recommendation Service for Mashup Creation

A Practical Cloud API Complementary Recommendation Service f...

引用

International Conference on computer Supported Cooperative Work in Design

作者： Xiaowei Liu Wenhui Chen Mengmeng Sun Yali Si Zhen Chen School of Information Science and Engineering Yanshan University Qinhuangdao China School of Computer Science and Engineering Changshu Institute of Technology Changshu China Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Yanshan University Qinhuangdao China

ISBN: (数字)9798350349184

ISBN: (纸本)9798350349191

With the popularity of the service-oriented development paradigm in the cloud era, numerous cloud APIs have been developed. The continuous proliferation of cloud APIs has submerged developers in a sea of options, bringing the need for cloud API recommendation techniques. Most existing cloud API recommendation methods focus on theoretical studies and neglect practical applications. As a result, these methods face several challenges, such as covering recommendation scenarios, generating real-time responses, and handling concurrent requests. In this paper, we propose and implement a practical cloud API complementary recommendation service for mashup creation. To address the above challenges, we first propose a cloud API complementary recommendation framework named FICR, which exploits feature interactions to mine implicit complementarity and models complementarity from the mashup side and the target cloud API side for scenario coverage. Then, we conduct performance optimization via precoding and batch size tuning to optimize FICR for real-time responses. Finally, we design a minimum-connection load balancing algorithm with resource early warning and build a distributed system for concurrent requests. Experiments conducted on real-world datasets show that FICR outperforms state-of-the-art baseline models in complementary recommendation scenarios, and demonstrate that the implemented service meets the real-time and concurrency requirements. The service will serve as a practical tool to help developers during mashup creation.

关键词： Concurrent computing Federated learning Mashups Precoding Load management Real-time systems Tuning

来源：评论

学校读者我要写书评

暂无评论

Performance analysis of synchronization buffering mechanism in peer-to-peer live streaming systems

引用

ICIC Express Letters, Part B: Applications 2012年第6期3卷 1665-1672页

作者： Jin, Shunfu Liu, Na Ma, Zhanyou College of Information Science and Engineering Yanshan University No. 438 Hebei Avenue Qinhuangdao 066004 China Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Yanshan University No. 438 Hebei Avenue Qinhuangdao 066004 China College of Science Yanshan University No. 438 Hebei Avenue Qinhuangdao 066004 China

In the peer-to-peer (P2P) live streaming systems, a single stream is decomposed into multiple sub-streams. For the participating nodes, data blocks of different sub-streams are stored in different sub-buffers for synchronization, then forwarded and assembled to the original video stream in chronological order. In this paper, regarding the sub-buffers as multiple queues with homogeneous Poisson arrival processes, the data block forwarding procedure as a polling service, the forwarding suspension in an empty sub-buffer as a service delay state, a symmetric polling model is constructed. By using the method of embedded Markov chain, the mean queueing length and the probability for one queue to be empty when it is visited are derived in steady-state. Moreover, the performance measures in terms of average synchronization buffering length and video continuity level are obtained accordingly. Finally, numerical results are given to demonstrate that an appropriate number of sub-streams and a reasonable time limit for forwarding suspension can enhance the system performance.© 2012 ISSN 2185-2766.

关键词： Synchronization

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：