<link rel="stylesheet" href="styles.cf390c6bfbe7555f.css">

學術出版;;Publications

Permanent URI for this communityhdl:11536/1

收集所有交通大學研究人員與學生個人或團體所發表之各項學術著作，涵蓋期刊論文、教師專書、學位論文等項。;;Collections of Publications of NCTU Researchers and Students Like Journal Papers and Thesis.

Browse

🏛️國立陽明交通大學機構典藏

Subject: search.filters.subject.Clustering

Search Results

Now showing 1 - 10 of 28
  • Thumbnail Image
    Item type:Item,
    Graceful Register Clustering by Effective Mean Shift Algorithm for Power and Timing Balancing
    (2019-01-01)
    Chang, Ya-Chu
    ;
    Lin, Tung-Wei
    ;
    Jiang, Iris Hui-Ru
    ;
    Nam, Gi-Joon
    ;
    電子工程學系及電子研究所
    As the wide adoption of FinFET technology in mass production, dynamic power becomes the bottleneck to achieving low power. Therefore, clock power reduction is crucial in modern IC design. Register clustering can effectively save clock power because of significantly reducing the number of clock sinks and register pin capacitance, clock routed wirelength, and the number of clock buffers. In this paper, we propose effective mean shift to naturally form clusters according to register distribution without placement disruption. Effective mean shift fulfills the requirements to be a good register clustering algorithm because it needs no prespecified number of clusters, is insensitive to initializations, is robust to outliers, is tolerant of various register distributions, is efficient and scalable, and balances clock power reduction against timing degradation. Experimental results show that our approach outperforms state-of-the-art work on power and timing balancing, as well as efficiency and scalability.
  • Thumbnail Image
    Item type:Item,
    A Generalized Enhanced Quantum Fuzzy Approach for Efficient Data Clustering
    (2019-01-01)
    Bharill, Neha
    ;
    Patel, Om Prakash
    ;
    Tiwari, Aruna
    ;
    Mu, Lifeng
    ;
    Li, Dong-Lin
    Data clustering is a challenging task to gain insights into data in various fields. In this paper, an Enhanced Quantum-Inspired Evolutionary Fuzzy C-Means (EQIE-FCM) algorithm is proposed for data clustering. In the EQIE-FCM, quantum computing concept is utilized in combination with the FCM algorithm to improve the clustering process by evolving the clustering parameters. The improvement in the clustering process leads to improvement in the quality of clustering results. To validate the quality of clustering results achieved by the proposed EQIE-FCM approach, its performance is compared with the other quantum-based fuzzy clustering approaches and also with other evolutionary clustering approaches. To evaluate the performance of these approaches, extensive experiments are being carried out on various benchmark datasets and on the protein database that comprises of four superfamilies. The results indicate that the proposed EQIE-FCM approach finds the optimal value of fitness function and the fuzzifier parameter for the reported datasets. In addition to this, the proposed EQIE-FCM approach also finds the optimal number of clusters and more accurate location of initial cluster centers for these benchmark datasets. Thus, it can be regarded as a more efficient approach for data clustering.
  • Thumbnail Image
    Item type:Item,
    Clustering data with partial background information
    (2019-05-01)
    Liu, Chien-Liang
    ;
    Hsaio, Wen-Hoar
    ;
    Chang, Tao-Hsing
    ;
    Li, Hsuan-Hsun
    ;
    資訊工程學系
    Clustering with partial supervision background information or semi-supervised clustering, learning from a combination of both labeled and unlabeled data, has received a lot of attention over the last decade. The supervisory information is usually used as the constraints to bias clustering towards a good region of search space. This paper proposes a semi-supervised algorithm, called constrained non-negative matrix factorization (Constrained-NMF), with a few labeled examples as constraints to improve performance. The proposed algorithm is a matrix factorization algorithm, in which initialization of matrices is required at the beginning. Although the benefits of good initialization are well-known, randomized seeding of basis matrix and coefficient matrix is still the standard approach for many non-negative matrix factorization (NMF) algorithms. This work devises an algorithm called entropy-based weighted semi-supervised fuzzy c-means (EWSS-FCM) algorithm to initialize the seeds. The experimental results indicate that the proposed Constrained-NMF can benefit from the initialization obtained from EWSS-FCM, which emphasizes the role of labeled examples and automatically weights them during the course of clustering. This work considers labeled examples in the objective functions to devise the two algorithms, in which the labeled information is propagated to unlabeled examples iteratively. We further analyze the proposed Constrained-NMF and give convergence justifications. The experiments are conducted on five real data sets, and experimental results indicate that the proposed algorithm generally outperforms the other alternatives.
  • Thumbnail Image
    Item type:Item,
    Automatic Clustering and Semantic Annotation for Dynamic IoT Sensor Data
    (2018-01-01)
    Yu, Ching-Tzu
    ;
    Zou, Yu-Hui
    ;
    Li, Hao-Yu
    ;
    Lin, Szu-Yin
    ;
    資訊管理與財務金融系 註：原資管所+財金所
    In a dynamic IoT environment, distributed sensors are used to collect real-time data continually. However, it is difficult to transform the dynamic data into a machine-readable and machine-interpretable form. we propose a semantic annotation approach to annotate sensor data via semantics. Firstly, this approach builds an ontology based on Semantic Sensor Network Ontology (SSN Ontology) for dynamic IoT sensor data. Then, the new knowledge is collected from input data by using the K-Means clustering, and to update the semantic information into the base ontology. The updated ontology forms the basis for semantic annotation.
  • Thumbnail Image
    Item type:Item,
    Generation of PUP-Keys on FPGAs by K-means Frequency Clustering
    (2018-01-01)
    Asha, K. A.
    ;
    Patyal, Abhishek
    ;
    Chen, Hung-Ming
    ;
    交大名義發表
    ;
    電子工程學系及電子研究所
    Physical Undonable functions (PUFs) generate secret keys that are unique and derived from the manufacturing process variations of the device. For high-security applications, generating the keys on the device is more secure than storing it on the device due to its resistance to attacks like tampering, probing and side-channel attacks. Also, the PUF-key generated should be very unique and reliable. In this paper, we are proposing a novel technique of generating a Ring Oscillator (RO) based PUF-keys on FPGAs by K-means clustering of RO frequencies. The placement of ROs on the FPGA fabric will have an impact on the variation of their oscillating frequencies. Based on the similarities in their frequency variations, the RO frequencies are grouped into k-clusters by K-means clustering and, the comparison of within-cluster RO frequencies will generate the PUF-keys. Our proposed method is evaluated on a large population of 193 Xilinx Spartan FPGAs. With the proposed method, the average inter die Ramming Weight of the PUF responses is 48.92% which is close to the ideal value of 50%, the average Ramming Weight of the n-bit PUF response is 49.90% and the average error rate due to changes in temperature and voltage is 3.43%.
  • Thumbnail Image
    Item type:Item,
    Personalized channel recommendation on live streaming platforms
    (2019-01-01)
    Lin, Chen-Yi
    ;
    Chen, Han-Shen
    ;
    資訊工程學系
    ;
    Department of Computer Science
    With unceasing technological advancements, an increasing number of viewers are watching channels through live streaming platforms, and live streaming technologies are developing rapidly. However, as thousands of channels are broadcasting on live streaming platforms, it is difficult for viewers to find their favorite channels. As a result, an accurate channel recommendation technique is required for the viewers. The current method of promoting live streaming channels recommends the most popular channels to viewers, but this ignores viewers' personal preferences. Therefore, we cluster viewers based on their personal preferences so that one cluster of viewers contains the viewers with similar favorite channels. In this way, the channels liked by viewers can be recommended to other viewers in the same group. In addition, our recommendation technique also considers viewers' loyalty towards a particular channel. In the experiment, a currently popular live streaming gaming platform, Twitch, is used for the analysis. The results confirm that our proposed recommendation technique is more accurate than the existing recommendation techniques.
  • Thumbnail Image
    Item type:Item,
    A Disk-Based Mining Algorithm for Frequent Pattern Discovery from Big Data in Distributed Computing Environments
    (2016-11-01)
    Lin, Kawuu W.
    ;
    Chung, Sheng-Hao
    ;
    Hsiao, Chun-Yuan
    ;
    Lin, Chun-Cheng
    ;
    Chen, Pei-Ling
    In distributed computing environments, frequent pattern mining by a multi-computing node can greatly improve mining efficiency. However, the drawback of memory limitations may cause interruption in the kernel and computing nodes when recursively building a frequent pattern (FP) tree or an FP-growth algorithm. In this paper, we propose disk-based FP-tree generation and node-based clustering mechanisms to solve the insufficient memory problem. Results from empirical evaluations show that the proposed method delivers excellent scalability.
  • Thumbnail Image
    Item type:Item,
    A Disk-Based Mining Algorithm for Frequent Pattern Discovery from Big Data in Distributed Computing Environments
    (2016-11)
    Lin, Kawuu W.
    ;
    Chung, Sheng-Hao
    ;
    Hsiao, Chun-Yuan
    ;
    Lin, Chun-Cheng
    ;
    Chen, Pei-Ling
    In distributed computing environments, frequent pattern mining by a multi-computing node can greatly improve mining efficiency. However, the drawback of memory limitations may cause interruption in the kernel and computing nodes when recursively building a frequent pattern (FP) tree or an FP-growth algorithm. In this paper, we propose disk-based FP-tree generation and node-based clustering mechanisms to solve the insufficient memory problem. Results from empirical evaluations show that the proposed method delivers excellent scalability.
  • Thumbnail Image
    Item type:Item,
    A Learning Scheme to Fuzzy C-Means based on a Compromise in Updating Membership Degrees
    (2014-01-01)
    Wu, Shang-Lin
    ;
    Lin, Yang-Yin
    ;
    Liu, Yu-Ting
    ;
    Chen, Chih-Yu
    ;
    Lin, Chin-Teng
    Fuzzy C-Means (FCM) clustering is the most well-known clustering method according to fuzzy partition for pattern classification. However, there are some disadvantages of using that clustering method, such as computational complexity and execution time. Therefore, to solve these drawbacks of FCM, the two-phase FCM procedure has been proposed in this study. Compared with the conventional FCM, the usage of a compromised learning scheme makes more adaptive and effective. By performing the proposed approach, the unknown data could be rapidly clustered according to the previous information. A synthetic data set with two dimensional variables is generated to estimate the performance of the proposed method, and to further demonstrate that our method not only reduces computational complexity but economizes execution time compared with the conventional FCM in each example.
  • Thumbnail Image
    Item type:Item,
    異質網路環境之行動搜尋關鍵技術---子計畫一：車用隨意網路存取控制與連結機制之研究
    (2008)
    簡榮宏
    ;
    JAN RONG-HONG
    ;
    國立交通大學資訊工程學系(所)
    隨著無線通訊網路技術的進步，加上電子元件價格漸漸下降以及行車安全問題越來越受重視，各國政府紛紛投入智慧型運輸系統(Intelligent Transport Systems, ITS)的研究。車用隨意網路(Vehicle Ad-hoc Network, VANET)，為當前了為實現ITS所發展出的重要網路架構。在個架構下，每一個搭載無線通訊設備的車輛，都可以透過路旁的路側系統(Road Side Unit, RSU) 連線到Server 端索取所需的資訊。而當車輛距離基地台太遠超出傳訊範圍時，也可透過其它的車輛幫忙轉送。換句話說，每一個車輛都可視為是一部行動無線路由器(mobile wireless router)。因此，這樣的架構可大幅提升網路建置彈性，只要有車輛的地方，即可是VANET的涵蓋範圍。 子計畫一的主要研究目標在於發展VANET網路之存取控制協定(Media Access Control, MAC)及連結機制(Connectivity Support)。分為三個階段:在第一年的計畫中，先針對網路低層的MAC進行設計，我們將基於未來最有可能被廣泛使用的專用短距離通通訊(Dedicated Short Range Communication , DSRC)標準進行發展，以合乎趨勢的走向。接下來，在第二年的研究裡，我們將為高速移動的車輛設計以有效的叢集化及功率指定方法，以減少網路干擾，進而提升流量產出及管理上的效果。基於這些低層的技術及規劃方法後，最後一年，我們將發展上層的路由及配送方法，以可克服VANET網路因高度移動和短距離傳輸所造成間歇性連結特性及其衍伸的延遲問題，達到可靠傳輸的最終目標。總括而言，我們將在三年中階段逐步完成下列三大主題: