检索结果-内蒙古大学图书馆

Extended Cross-Modality United Learning for Unsupervised Visible-Infrared Person Re-Identification

IEEE SIGNAL PROCESSING LETTERS 2025年 32卷 2154-2158页

作者： Wu, Ruixing Yang, Yiming He, Jiakai Hu, Haifeng Sun Yat Sen Univ Sch Elect & Informat Technol Guangzhou 510006 Peoples R China Chinese Univ Hong Kong Shenzhen Sch Sci & Engn Shenzhen 518172 Peoples R China

Unsupervised learning visible-infrared person re-identification (USL-VI-ReID) aims to learn modality-invariant features from unlabeled cross-modality data. However, existing approaches lack comprehensive cross-modality clustering or excessively pursue cluster-level association, which hinders reliable learning of modality-invariant features. To address these challenges, we propose an Extended Cross-Modality United Learning (ECUL) framework, which integrates Extended Modality-Camera clustering (EMCC) and Two-Step Memory Updating Strategy (TSMem) modules. Specifically, we design ECUL to naturally unify intra-modality clustering, inter-modality clustering, and inter-modality instance selection, establishing compact and accurate cross-modality associations while reducing the introduction of noisy labels. Moreover, EMCC captures and filters neighborhood relationships by extending the encoding vector, which further promotes the learning of modality-invariant and camera-invariant knowledge in terms of the clustering algorithm. Finally, TSMem provides accurate and generalized proxy points for contrastive learning by updating memory in stages. Comprehensive experiments conducted on the SYSU-MM01 and RegDB datasets demonstrate that the proposed ECUL framework shows promising performance and even outperforms certain supervised methods.

关键词： Training Cameras Encoding Accuracy Contrastive learning clustering algorithms Vectors Noise Signal processing algorithms Noise measurement Person re-identification (Re-ID) clustering unsupervised learning visible-infrared

来源：评论

学校读者我要写书评

暂无评论

LDCA: Lightweight Dynamic clustering Algorithm for IoT-Connected Wide-Area WSN and Mobile Data Sink Using LoRa

引用

IEEE INTERNET OF THINGS JOURNAL 2022年第2期9卷 1313-1325页

作者： Rahman, Gazi M. E. Wahid, Khan A. Univ Saskatchewan Dept Elect & Comp Engn Saskatoon SK S7N 5A9 Canada

Wide-area monitoring applications of the Internet-of-Things (IoT) connected wireless sensor network (WSN) consists of sensor nodes (SNs) with limited hardware and energy sources. The distributed nature of such a network and the difficulty of remote access make it more demanding to design an energy-efficient WSN. Moreover, long-range and low-power wireless connectivity is a challenge in IoT-connected applications. Present WSN topologies deal mainly with fixed SNs, SN distribution, and fixed data sink (DS). The majority of the control layer is implemented in the lower hierarchical layer or in a virtual middle layer, which reduces the network lifetime due to excessive processing and data transmission activities. This article proposes a real-time lightweight dynamic clustering algorithm (LDCA) for a WSN that supports the following two scenarios with limited processing resources: 1) with mobile DSs and static SNs (such as DS and SNs mounted on unmanned aerial vehicles, autonomous vehicles) and 2) with mobile SNs and static DSs (such as livestock monitoring or autonomous robots in smart farming, and urban monitoring). The proposed algorithm is based on the received signal strength indicator and signal-to-noise ratio of a long-range (LoRa) interface and its residual energy. Mathematical models were derived for real-time clustering using LoRa. Memory requirement and clustering efficiency of the constrained SN and DS for various mobility scenario were evaluated. The proposed LDCA reduces the energy requirement to 33% compared to static clustering, by reducing the number of concurrent clusters and hops. In addition, a hardware-based approach was used to validate the LDCA algorithm, and evaluate its performance in terms of energy efficiency, packet delivery rate, and network lifetime.

关键词： Wireless sensor networks Heuristic algorithms clustering algorithms Monitoring Wireless communication Topology Network topology Dynamic clustering Internet of Things (IoT) lightweight algorithm long-range (LoRa) mobile sensor node (SN) mobile sink wireless sensor network (WSN)

来源：评论

学校读者我要写书评

暂无评论

A Novel Kernelized Total Bregman Divergence-Driven Possibilistic Fuzzy clustering With Multiple Information Constraints for Image Segmentation

引用

IEEE TRANSACTIONS ON FUZZY SYSTEMS 2022年第6期30卷 1624-1639页

作者： Wu, Chengmao Zhang, Xue Xian Univ Posts & Telecommun Sch Elect Engn Xian 710121 Peoples R China

Aiming at the problem that existing robust fuzzy clustering algorithms are still sensitive to high noise, a total Bregman divergence-driven possibilistic fuzzy clustering with multiple information constraints and kernel metric (TSKFLICM) is proposed in this article. First, we introduce total Bregman divergence (TBD) to overcome the shortcoming that Bregman divergence is variant with rotation. Second, polynomial kernel function is used to kernelize TBD, and kernelized TBD is embedded with neighborhood information of pixels to further enhance its ability to suppress noise. Finally, kernelized TBD with spatial information constraints is combined with possibilistic typicality to construct the novel objective function of possibilistic fuzzy clustering, and a TBD-driven kernel possibilistic fuzzy clustering with multiple information constraints is obtained through optimization theory. The effectiveness of the proposed algorithm for noisy image segmentation is explained by means of sample weighted clustering method. Experimental results show that compared with other algorithms, the Jaccard score, segmentation accuracy, and peak signal-to-noise ratio of the proposed algorithm are improved by 0.017-0.481, 1.327%-41.260%, and 2.416-11.765, respectively. Therefore, the TSKFLICM algorithm has better segmentation performance and stronger antinoise ability than the existing state-of-the-art fuzzy clustering algorithm.

关键词： clustering algorithms Image segmentation Kernel Euclidean distance Robustness Linear programming Noise measurement Fuzzy clustering image segmentation polynomial kernel possibilistic clustering spatial information constraints total Bregman divergence (TBD)

来源：评论

学校读者我要写书评

暂无评论

SCHAIN-IRAM: An Efficient and Effective Semi-Supervised clustering Algorithm for Attributed Heterogeneous Information Networks

引用

IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING 2022年第4期34卷 1980-1992页

作者： Li, Xiang Wu, Yao Ester, Martin Kao, Ben Wang, Xin Zheng, Yudian Univ Hong Kong Dept Comp Sci Hong Kong Peoples R China Twitter San Francisco CA 94103 USA Simon Fraser Univ Sch Comp Sci Burnaby BC V5A 1S6 Canada Tsinghua Univ Dept Comp Sci & Technol Beijing 100084 Peoples R China

A heterogeneous information network (HIN) is one whose nodes model objects of different types and whose links model objects' relationships. To enrich its information, objects in an HIN are typically associated with additional attributes. We call such an HIN an Attributed HIN or AHIN. We study the problem of clustering objects in an AHIN, taking into account objects' similarities with respect to both object attribute values and their structural connectedness in the network. We show how supervision signal, expressed in the form of a must-link set and a cannot-link set, can be leveraged to improve clustering results. We put forward the SCHAIN algorithm to solve the clustering problem, and two highly efficient variants, SCHAIN-PI and SCHAIN-IRAM, which employ the power iteration based method and the implicitly restarted Arnoldi method respectively to compute eigenvectors of a matrix. We conduct extensive experiments comparing SCHAIN-based algorithms with other state-of-the-art clustering algorithms. Our results show that SCHAIN-IRAM outperforms other competitors in terms of clustering effectiveness and is highly efficient.

关键词： clustering algorithms Task analysis Facebook Data mining Linear programming Data models Computer science Semi-supervised clustering attributed heterogeneous information network object attributes network structure

来源：评论

学校读者我要写书评

暂无评论

Learnable Weighting of Intra-Attribute Distances for Categorical Data clustering with Nominal and Ordinal Attributes

引用

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 2022年第7期44卷 3560-3576页

作者： Zhang, Yiqun Cheung, Yiu-ming Guangdong Univ Technol Sch Comp Guangzhou 510080 Peoples R China Hong Kong Baptist Univ Dept Comp Sci Hong Kong Peoples R China

The success of categorical data clustering generally much relies on the distance metric that measures the dissimilarity degree between two objects. However, most of the existing clustering methods treat the two categorical subtypes, i.e., nominal and ordinal attributes, in the same way when calculating the dissimilarity without considering the relative order information of the ordinal values. Moreover, there would exist interdependence among the nominal and ordinal attributes, which is worth exploring for indicating the dissimilarity. This paper will therefore study the intrinsic difference and connection of nominal and ordinal attribute values from a perspective akin to the graph. Accordingly, we propose a novel distance metric to measure the intra-attribute distances of nominal and ordinal attributes in a unified way, meanwhile preserving the order relationship among ordinal values. Subsequently, we propose a new clustering algorithm to make the learning of intra-attribute distance weights and partitions of data objects into a single learning paradigm rather than two separate steps, whereby circumventing a suboptimal solution. Experiments show the efficacy of the proposed algorithm in comparison with the existing counterparts.

关键词： clustering algorithms Weight measurement Measurement Loss measurement Encoding Task analysis Partitioning algorithms Categorical data clustering nominal-and-ordinal attribute intra-attribute distance learnable weighting

来源：评论

学校读者我要写书评

暂无评论

Feature Saliencies in Asymmetric Hidden Markov Models

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2024年第3期35卷 3586-3600页

作者： Puerto-Santana, Carlos Larranaga, Pedro Bielza, Concha Aingura IIoT Donostia San Sebastian 20009 Spain Univ Politecn Madrid Dept Inteligencia Artificial ETS Ingenieros Informat Madrid 28660 Spain

Many real-life problems are stated as nonlabeled high-dimensional data. Current strategies to select features are mainly focused on labeled data, which reduces the options to select relevant features for unsupervised problems, such as clustering. Recently, feature saliency models have been introduced and developed as clustering models to select and detect relevant variables/features as the model is learned. Usually, these models assume that all variables are independent, which narrows their applicability. This article introduces asymmetric hidden Markov models with feature saliencies, i.e., models capable of simultaneously determining during their learning phase relevant variables/features and probabilistic relationships between variables. The proposed models are compared with other state-of-the-art approaches using synthetic data and real data related to grammatical face videos and wear in ball bearings. We show that the proposed models have better or equal fitness than other state-of-the-art models and provide further data insights.

关键词： Hidden Markov models Data models Computational modeling Bayes methods Markov processes Feature extraction clustering algorithms Asymmetric information Bayesian networks embedded feature selection feature saliencies feature subset selection (FSS) hidden Markov models (HMMs)

来源：评论

学校读者我要写书评

暂无评论

Outlier Detection Based On Heuristic clustering Algorithm 7

Outlier Detection Based On Heuristic Clustering Algorithm

引用

7th IEEE International Conference on Electronic Information and Communication Technology, ICEICT 2024

作者： Xue, Mingyang Wen, Ruihu Li, Jiabao Xi'an Institute of Electromechanical Information Technology Xi'an China

ISBN: (纸本)9798350384437

Outlier detection is a basic problem in machine learning, which is widely used in industrial anomaly detection, network performance monitoring and activity monitoring. Most of the outlier detection algorithms are not effective in local outlier recognition, and can not distinguish the group points from the inner points in the data set with high outlier proportion. Moreover, the effectiveness of the algorithm is largely affected by the density and shape of the data set. This paper presents an outlier detection method based on heuristic clustering algorithm. The algorithm is divided into three stages. In the first stage, a heuristic clustering algorithm based on centroid learning is used to cluster the original data. In the second stage, the outlier score is calculated by the contraction distance of the target point, and the initial outlier identification is carried out. In the third stage, the "backward"mechanism is used to eliminate the "wrong points". The algorithm iterates through these three stages until the stop condition is met. In short, we propose a new outlier detection algorithm based on the heuristic clustering algorithm framework of centroid learning, and propose a "backward"mechanism to correct the wrong or wrong points. Finally, the validity of the proposed outlier detection algorithm is verified on both synthetic and real data sets. © 2024 IEEE.

关键词： clustering algorithms

来源：评论

学校读者我要写书评

暂无评论

Raft Consensus Grouping Mechanism Based on Affinity Propagation clustering Algorithm 4

Raft Consensus Grouping Mechanism Based on Affinity Propagat...

引用

4th International Conference on Consumer Electronics and Computer Engineering, ICCECE 2024

作者： Zuo, Nan Chen, Yubin Zheng, Yuanjie School of Software Nanchang Hangkong University Nanchang China

ISBN: (纸本)9798350384062

An approach to address the shortcomings in the performance of existing consensus algorithms is the proposal of a Raft consensus grouping mechanism, which is based on the nearest neighbor propagation clustering algorithm: AP-Raft. First, the input eigenvalues of the algorithm are gathered from the network transmission efficiency and the current memory usage rate of each node. Second, the nearest neighbor propagation clustering technique groups all nodes based on the feature values. This allows each group's memory consumption rate and node network transmission efficiency to be approximated within a range. Finally, a leader node is chosen for each group, and to efficiently increase the load among the leader nodes, each leader node is connected using a consistent hash technique. Following experimental research, the AP-Raft method successfully increases the system's throughput and consensus time delay while maintaining data consistency when compared to the conventional Raft algorithm. © 2024 IEEE.

关键词： clustering algorithms

来源：评论

学校读者我要写书评

暂无评论

A Multiview clustering Method With Low-Rank and Sparsity Constraints for Cancer Subtyping

引用

IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS 2022年第6期19卷 3213-3223页

作者： Huang Zhanpeng Wu Jiekang Guangdong Univ Technol Sch Automat Guangzhou 510006 Guangdong Peoples R China Univ Guangdong Prov Med Informat & Real World Engn Technol Ctr Guangzhou 510006 Guangdong Peoples R China

Multiomics data clustering is one of the major challenges in the field of precision medicine. Integration of multiomics data for cancer subtyping can improve the understanding on cancer and reveal systems-level insights. How to integrate multiomics data for accurate cancer subtyping is an interesting and challenging research problem. To capture the global and the local structure of omics data, a novel framework for integrating multiomics data is proposed for cancer subtyping. Multiview clustering with low-rank and sparsity constraints (MVCLRS) can measure the local similarities of samples in each omics data and obtain global consensus structures by integrating the multiomics data. The main insight provided by MVCLRS is that low-rank sparse subspace clustering for the construction of an affinity matrix can best capture the local similarities in omics data. Extensive testing is conducted on 10 real world cancer datasets with multiomics from The Cancer Genome Atlas. Compared with 10 state-of-the-art multiomics clustering algorithms, the MVCLRS performs better in the 10 cancer datasets by providing its clustering results with at least one enriched clinical label in nine of ten cancer subtypes, the most of any method.

关键词： clustering algorithms Cancer Data models Biology Sparse matrices Minimization Analytical models Multiview clustering subspace clustering low-rank representation sparsity cancer subtyping

来源：评论

学校读者我要写书评

暂无评论

English Automatic Teaching Intelligent Platform Based on Improved Dynamic Time Warping Algorithm

引用

IEIE Transactions on Smart Processing and Computing 2025年第2期14卷 165-177页

作者： Zhang, Qianwen School of Education Jingzhou University Jingzhou434000 China

Modern educational concepts are student-centered, emphasizing personalized and comprehensive development of students. However, traditional English teaching methods have problems such as single teaching content and insufficient teacher resources. Based on the dynamic time warping, this study introduces clustering algorithm to construct an automatic intelligent platform for English teaching. On the basis of data analysis and clustering, the constructed English automatic teaching intelligent platform is evaluated. Experiments showed that the proposed algorithm could effectively improve the weaknesses of traditional algorithms. There was no crossing between data points. The clustering effect of data points was significant. The accuracy is the highest at 0.91 under different classification point splits. The research system platform had the highest correlation coefficient for speech recognition, ranging from 0.7 to 0.8. Its center frequency was within 1500. The center frequency variance was within 200. When the number of knowledge points was 50, the average teaching time using the proposed teaching system was reduced by nearly 300 minutes. In summary, the English automatic teaching intelligent platform can solve the problems existing in traditional English teaching methods, improving the effectiveness of English teaching. © 2025 Institute of Electronics Engineers of Korea. All rights reserved.

关键词： clustering algorithms

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：