Minimum Number Of Clusters Research Articles

Surface Mount Technology Line Optimisation using Modified k-means with Feature Weight Constraints

The common setup problem for chip mounters in a Surface Mount Technology (SMT) line is to group different device models, each of them consisting of different components, into minimum number of clusters where each cluster has a maximum component size. This type of setup problem is classified as a clustering problem with feature weight constraints which cannot be fulfilled by traditional data clustering. In this study, we introduce some vital modifications on standard k-means algorithm such that it can incorporate feature weight constraints adapted from cluster size constraints. We also propose a modification to the elbow method to determine the number of clusters of the clustering problem with feature weight constraints. The results show that the proposed algorithm (modified k-means with feature weight constraints) is able to fulfill the feature weight constraints and solve the common setup problem. The results verify the proposed algorithm has superior performance over standard k-means algorithm in clustering problem with feature weight constraints. For the common setup problem on a given data set, the analysis shows that 1000 runs of the proposed algorithm implemented using MATLAB is able to obtain at least one valid clustering result within a reasonable run time.

A novel protocol for stable period and lifetime enhancement in WSN

Energy optimization is one of the open issues in wireless sensor networks. Most of the work focuses on clustering techniques to reduce the energy consumption and enhancement of stability period. In this paper we have proposed a novel region based routing protocol for the rise in stability duration and lifetime improvement. The whole sensing area is divided among nine regions and dedicated to different variety of nodes. Dynamically changing cluster head election probability has been exploited for heterogeneous wireless sensor networks. In this paper, cluster head selection is based on residual energy of nodes, region in which the node is residing and minimum number of clusters per round. The proposed protocol is simulated using MATLAB and the results establish that it performs better than the ETSSEP in terms of stability period, throughput and lifetime. Proposed routing protocol frames more stable routing environment than the ETSSEP. It is also evident from the results that stability is increased by 35.03% in comparison to ETSSEP. Lifetime and throughput have also been improved by 23.16% (for 50% live node) and 25.58% respectively.

Design and Comparative Analysis of New Personalized Recommender Algorithms with Specific Features for Large Scale Datasets

Nowadays, because of the tremendous amount of information that humans and machines produce every day, it has become increasingly hard to choose the more relevant content across a broad range of choices. This research focuses on the design of two different intelligent optimization methods using Artificial Intelligence and Machine Learning for real-life applications that are used to improve the process of generation of recommenders. In the first method, the modified cluster based intelligent collaborative filtering is applied with the sequential clustering that operates on the values of dataset, user′s neighborhood set, and the size of the recommendation list. This strategy splits the given data set into different subsets or clusters and the recommendation list is extracted from each group for constructing the better recommendation list. In the second method, the specific features-based customized recommender that works in the training and recommendation steps by applying the split and conquer strategy on the problem datasets, which are clustered into a minimum number of clusters and the better recommendation list, is created among all the clusters. This strategy automatically tunes the tuning parameter λ that serves the role of supervised learning in generating the better recommendation list for the large datasets. The quality of the proposed recommenders for some of the large scale datasets is improved compared to some of the well-known existing methods. The proposed methods work well when λ = 0.5 with the size of the recommendation list, |L| = 30 and the size of the neighborhood, |S| < 30. For a large value of |S|, the significant difference of the root mean square error becomes smaller in the proposed methods. For large scale datasets, simulation of the proposed methods when varying the user sizes and when the user size exceeds 500, the experimental results show that better values of the metrics are obtained and the proposed method 2 performs better than proposed method 1. The significant differences are obtained in these methods because the structure of computation of the methods depends on the number of user attributes, λ, the number of bipartite graph edges, and |L|. The better values of the (Precision, Recall) metrics obtained with size as 3000 for the large scale Book-Crossing dataset in the proposed methods are (0.0004, 0.0042) and (0.0004, 0.0046) respectively. The average computational time of the proposed methods takes <10 seconds for the large scale datasets and yields better performance compared to the well-known existing methods.

Tree-based text stream clustering with application to spam mail classification

This paper proposes a new text clustering algorithm based on a tree structure. The main idea of the clustering algorithm is a sub-tree at a specific node represents a document cluster. Our clustering algorithm is a single pass scanning algorithm which traverses down the tree to search for all clusters without having to predefine the number of clusters. Thus, it fits our objectives to produce document clusters having high cohesion, and to keep the minimum number of clusters. Moreover, an incremental learning process will perform after a new document is inserted into the tree, and the clusters will be rebuilt to accommodate the new information. In addition, we applied the proposed clustering algorithm to spam mail classification and the experimental results show that tree-based text clustering spam filter gives higher accuracy and specificity than the cobweb clustering, naive Bayes and KNN.

A Novel Approximation for Multi-Hop Connected Clustering Problem in Wireless Networks

Wireless sensor networks (WSNs) have been widely used in a plenty of applications. To achieve higher efficiency for data collection, WSNs are often partitioned into several disjointed clusters, each with a representative cluster head in charge of the data gathering and routing process. Such a partition is balanced and effective, if the distance between each node and its cluster head can be bounded within a constant number of hops, and any two cluster heads are connected. Finding such a cluster partition with minimum number of clusters and connectors between cluster heads is defined as minimum connected d-hop dominating set (d-MCDS) problem, which is proved to be NP-complete. In this paper, we propose a distributed approximation named CS-Cluster to address the d-MCDS problem under unit disk graph. CS-Cluster constructs a sparser d-hop maximal independent set (d-MIS), connects the d-MIS, and finally checks and removes redundant nodes. We prove the approximation ratio of CS-Cluster is (2d + 1)λ, where λ is a parameter related with d but is no more than 18.4. Compared with the previous best result O(d <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ), our approximation ratio is a great improvement. Our evaluation results demonstrate the outstanding performance of our algorithm compared with previous works.

Energy Efficient Genetic Inspired Scheduling for Data Aggregation

Energy efficient Data Aggregation Scheduling is essential in Ubiquitous sensor networks. Optimal communication distances between the nodes and the base station improves the network lifetime. Clustering improves the efficiency of the data aggregation problem. Many clustering techniques exist to find the optimal number of clusters in the network. The computation complexity of the methods to obtain optimal number of clusters is fair when genetic approaches are employed instead of the classical approaches. Since nodes in USN are dynamic, finding data aggregation schedules is difficult in a USN. Concurrent transmissions improve the throughput of the network, but the SINR perceived at the receiver should be greater than or equal to a certain threshold value for a successful transmission. Also the cluster heads and the member nodes dissipate energy. This power dissipation is more when there are more number of cluster heads. The noise and the power dissipation play a major role in decreasing the network life time. Hence in this paper we designed an efficient energy dissipation algorithm for data aggregation scheduling using the principles of Genetic algorithm and SINR. Since the battery power and bandwidth are the limited resources for the nodes in the USN our data aggregation scheduling algorithm gives equal priority for all of the following while electing a node as a cluster head. They are (a) Total power dissipation from all cluster heads in the USN, (b) The total power dissipation from the nodes at each cluster heads, (c) The SINR perceived at the cluster head should be more than a threshold value at each cluster head for the transmission to be successful, (d) The optimal schedule for the nodes in the USN so that all the nodes transmits their data finally to the base station quickly. With the application of the genetic methods our algorithm proved to be efficient when compared with the existing algorithms in obtaining maximum network life time with minimum number of clusters.

Competitive self-stabilizing k-clustering

In this paper, we give a silent self-stabilizing algorithm for constructing a k-clustering of any asynchronous connected network with unique IDs. Our algorithm stabilizes in O(n) rounds, using O(log⁡k+log⁡n) space per process, where n is the number of processes. In the general case, our algorithm constructs O(nk)k-clusters. If the network is a Unit Disk Graph (UDG), then our algorithm is 7.2552k+O(1)-competitive, that is, the number of k-clusters constructed by the algorithm is at most 7.2552k+O(1) times the minimum possible number of k-clusters in any k-clustering of the same network. More generally, if the network is an Quasi-Unit Disk Graph (QUDG) with approximation ratio λ, then our algorithm is 7.2552λ2k+O(λ)-competitive. In case of tree networks, our algorithm computes a k-clustering with the minimum number of clusters. Our solution is based on the self-stabilizing construction of a data structure called an MIS tree, a spanning tree of the network whose processes at even levels form a maximal independent set of the network. The MIS tree construction we use (called LFMIS) is the time bottleneck of our k-clustering algorithm, as it takes Θ(n) rounds in the worst case, while the rest of the algorithm takes O(D) rounds, where D is the diameter of the network. We would like to improve that time to be O(D), but we show that our distributed MIS tree construction is a P-complete problem.

A novel weighted clustering algorithm in mobile ad hoc networks using discrete particle swarm optimization (DPSOWCA)

AbstractIn this paper, a novel weighted clustering algorithm in mobile ad hoc networks using discrete particle swarm optimization (DPSOWCA) is proposed. The proposed algorithm shows how discrete particle swarm optimization can be useful in enhancing the performance of clustering algorithms in mobile ad hoc networks. Consequently, it results in the minimum number of clusters and hence minimum cluster heads. The goals of the algorithm are to minimize the number of cluster heads, to enhance network stability, to maximize network lifetime, and to achieve good end‐to‐end performance. Analysis and simulation of the algorithm have been implemented and the validity of the algorithm has been proved. Results show that the proposed algorithm performs better than the existing weight‐based clustering algorithm and adapts to different kinds of network conditions. Copyright © 2009 John Wiley &amp; Sons, Ltd.

SU‐FF‐I‐79: Evaluation of Reproducibility of FMRI Maps in Patients with Proven Low‐Grade Brain Neoplasms

Purpose: The focus of this project is to demonstrate the test re‐test reliability of fMRI in subjects with benign and biopsy proven low‐grade neoplasms. Methods and Materials: FMRI activation maps with different thresholds were generated for 6 subjects with benign and biopsy proven low‐grade brain neoplasms using a GE Signa 1.5T MRI scanner and birdcage head coil. The imaging protocol included a 3DSPGR T1 brain volume, a coronal T2 anatomical acquisition(24 slices), and 8 functional scans. The 8 functional scans consisted of four functional task repeated twice. Each task verified one of the following: language, language comprehension, fine motor control, and visual perception with Blood Oxygen Dependent(BOLD) responses. Preprocessing steps included dicom to afni conversion, motion correction and spatial smoothing(9mm Gaussian filter). A correlation analysis was performed which allowed the reference timing of the BOLD response in the expected areas to be confirmed. A measure of reproducibility was calculated using the Euclidean distance between the center of mass of the activated volumes in two independent scans and measuring the number of reproducible activated voxels, Rsize. Results and Discussion: Activated clusters were observed for each functional scan. Activated clusters were not observed in all subjects for each task. Rsize was calculated to measure the reproducibility. These measurements will vary between 0.0(worst) and 1.0(best). The Text Listening Task provided the best reproducibility of activated clusters with an average value of 0.94±0.05 and 0.90±0.06 for the left and right temporal lobe, respectively. Activation of clusters amongst subjects varied greatly for the visual perception scan. Previous published results had similar findings. Conclusions: Similar comparisons of reproducibility will be made for more subjects. Roverlap, another parameter used for testing reproducibility will be calculated. The minimum number of clusters used and the cluster radius will be further examined to obtain more activated clusters.

Locating cyclone shelters: a case

PurposeThis paper aims to present the cyclone shelter location problem as a problem of grouping a set of villages into a minimum number of clusters and finding the location and capacity of the shelters.Design/methodology/approachProposes an algorithm to solve the location problem that uses a distance matrix of the villages, applies Elzinga‐Hearn method iteratively to group them into circular clusters, and determines the shelter size. It is implemented in a software package, which reads the village records from a database, executes the steps of the algorithm, and writes the results into a file.FindingsThe villages are grouped into 13 clusters, eight requiring cyclone shelters to be built and five requiring no shelter, taking into consideration the available space in double‐storeyed buildings in the villages. The capacity and location of the each shelter are also obtained.Research limitations/implicationsThe approach can be modified to address several practical constraints. For example, for the villages situated far from the coast, the limiting travel distance can be increased, considering that a lesser intensity cyclone would give more time to the villagers to reach the shelter.Practical implicationsFunds are available from a number of public and private organizations to construct cyclone shelters, which could be used properly only when optimal locations for the cyclone shelters are identified.Originality/valueOriginality of the work lies in adapting Elzinga‐Hearn method iteratively to group the villages into circular clusters and implementing the algorithm in software solution. The concerned authority, to take cyclone shelter location decisions, can use the software.

Clustering Binary Oligonucleotide Fingerprint Vectors for DNA Clone Classification Analysis

We considered the problem of clustering binarized oligonucleotide fingerprints that attempts to identify clusters. Oligonucleotide fingerprinting is a powerful DNA array based method to characterize cDNA and rRNA libraries and has many applications including gene expression profiling and DNA clone classification. DNA clone classification is the main application for the problem considered in this paper. Most of the existing approaches for clustering use normalized real intensity values and thus do not treat positive and negative hybridization signals equally. This is demonstrated in a series of recent publications where a discrete approach typically useful in the classification of microbial rRNA clones has been proposed. In the discrete approach, hybridization intensities are normalized and thresholds are set such that a value of 1 represents hybridization, a value of 0 represents no hybridization, and an N represents unknown, which is also called a missing value. A combinatorial optimization problem is then formulated attempting to cluster the fingerprints and resolve the missing values simultaneously. It has been examined that missing values cause much difficulty in clustering analysis and most clustering methods are very sensitive to them. In this paper, we turned a little back to the traditional clustering problem, which takes in no missing values but with the revised goal to stabilize the number of clusters and maintain the clustering quality. We adopted the binarizing scheme used in the discrete approach as it is shown to be typically useful for the clone classifications. We formulated such a problem into another combinatorial optimization problem. The computational complexity of this new clustering problem and its relationships to the discrete approach and the traditional clustering problem were studied. We have designed an exact algorithm for the new clustering problem, which is an A* search algorithm for finding a minimum number of clusters. The experimental results on two commonly tested real datasets demonstrated that the A* search algorithm runs fast and performs better than some popular hierarchical clustering methods, in terms of separating clones that have different characteristics with respect to the given oligonucleotide probes.

Implementing multiple static field delivery for intensity modulated beams.

A clinically oriented two-dimensional intensity-modulated beam delivery method is implemented using multiple static segmented fields, i.e., the "step-and-shoot" approach. Starting with a desired al" intensity distribution, it creates a multiple-level intensity approximation, and then constructs a sequence of segmented fields to deliver the multiple-level intensities using multileaf collimator (MLC) and independent backup jaws. The approach starts with a simple grouping of all the nonzero intensity values into a minimum number of clusters for a user specified deviation tolerance for the ideal plan. The k-means clustering algorithm is then employed to find the optimal levels of intensity that minimize the discrepancies between the ideal and the approximated intensities, without violating the user specified deviation tolerance. The multiple-level intensities are then decomposed into a sequence of machine deliverable segments. Apart from the first segment for each gantry angle, all the other segments are arranged to minimize the total travel distance of the leaves. The first segment covers the entire irradiated area and is used for treatment verification by electronic portal imaging. The implementation issues due to the physical constraints of the MLCs are also addressed.

Simulated ROSAT observations of clusters of galaxies

Detailed simulations have been made of the search for clusters of galaxies during the ROSAT all-sky x-ray survey (0.1–2 keV). The simulations use a conservative estimate of the cluster luminosity function to create an x-ray catalog specifying cluster position, redshift, flux, temperature and angular extent. This is used in combination with realistic models of the ROSAT scanning motion and of the background due to particles, cosmic x-rays and scattered solar x-rays, to build a simulated ROSAT data base. The data base has been subjected to the ROSAT source search algorithms and the results analyzed. Two cases were considered, a 2-degree wide great circle strip representing the general survey, and a region within 10 degrees of the ecliptic pole, where the exposure is unusually high. We present estimates of the minimum number of clusters to be detected, the fraction recognized as extended x-ray sources, and the minimum detectable energy flux from clusters during the survey.