Mixed fuzzy inter-cluster separation clustering algorithm

Based on inter-cluster separation clustering (ICSC) fuzzy inter-cluster separation clustering (FICSC) deals with all the distances between the cluster centers, maximizes these distances and obtains the better performances of clustering. However, FICSC is sensitive to noises the same as fuzzy c-means (FCM) clustering. Possibilistic type of FICSC is proposed to combine FICSC and possibilistic c-means (PCM) clustering. Mixed fuzzy inter-cluster separation clustering (MFICSC) is presented to extend possibilistic type of FICSC because possibilistic type of FICSC is sensitive to initial cluster centers and always generates coincident clusters. MFICSC can produce both fuzzy membership values and typicality values simultaneously. MFICSC shows good performances in dealing with noisy data and overcoming the problem of coincident clusters. The experimental results with data sets show that our proposed MFICSC holds better clustering accuracy, little clustering time and the exact cluster centers.     

Automatic clustering using genetic algorithms

In face of the clustering problem, many clustering methods usually require the designer to provide the number of clusters as input. Unfortunately, the designer has no idea, in general, about this information beforehand. In this article, we develop a genetic algorithm based clustering method called automatic genetic clustering for unknown K (AGCUK). In the AGCUK algorithm, noising selection and division-absorption mutation are designed to keep a balance between selection pressure and population diversity. In addition, the Davies-Bouldin index is employed to measure the validity of clusters. Experimental results on artificial and real-life data sets are given to illustrate the effectiveness of the AGCUK algorithm in automatically evolving the number of clusters and providing the clustering partition.     

Enumerating K best paths in length order in DAGs

We address the problem of finding the K best paths connecting a given pair of nodes in a directed acyclic graph (DAG) with arbitrary lengths. One of the main results in this paper is the proof that a tree representing the kth shortest path is obtained by an arc exchange in one of the previous (k − 1) trees (each of which contains a previous best path). An O(m + K(n + log K)) time and O(K + m) space algorithm is designed to explicitly determine the K shortest paths in a DAG with n nodes and m arcs. The algorithm runs in O(m + Kn) time using O(K + m) space in DAGs with integer length arcs. Empirical results confirming the superior performance of the algorithm to others found in the literature for randomly generated graphs are reported.     

PSOHS: an efficient two-stage approach for data clustering

Cluster analysis is an important task in data mining and refers to group a set of objects such that the similarities among objects within the same group are maximal while similarities among objects from different groups are minimal. The particle swarm optimization algorithm (PSO) is one of the famous metaheuristic optimization algorithms, which has been successfully applied to solve the clustering problem. However, it has two major shortcomings. The PSO algorithm converges rapidly during the initial stages of the search process, but near global optimum, the convergence speed will become very slow. Moreover, it may get trapped in local optimum if the global best and local best values are equal to the particle’s position over a certain number of iterations. In this paper we hybridized the PSO with a heuristic search algorithm to overcome the shortcomings of the PSO algorithm. In the proposed algorithm, called PSOHS, the particle swarm optimization is used to produce an initial solution to the clustering problem and then a heuristic search algorithm is applied to improve the quality of this solution by searching around it. The superiority of the proposed PSOHS clustering method, as compared to other popular methods for clustering problem is established for seven benchmark and real datasets including Iris, Wine, Crude Oil, Cancer, CMC, Glass and Vowel.     

A Global Optimization RLT-based Approach for Solving the Hard Clustering Problem

The field of cluster analysis is primarily concerned with the sorting of data points into different clusters so as to optimize a certain criterion. Rapid advances in technology have made it possible to address clustering problems via optimization theory. In this paper, we present a global optimization algorithm to solve the hard clustering problem, where each data point is to be assigned to exactly one cluster. The hard clustering problem is formulated as a nonlinear program, for which a tight linear programming relaxation is constructed via the Reformulation-Linearization Technique (RLT) in concert with additional valid inequalities that serve to defeat the inherent symmetry in the problem. This construct is embedded within a specialized branch-and-bound algorithm to solve the problem to global optimality. Pertinent implementation issues that can enhance the efficiency of the branch-and-bound algorithm are also discussed. Computational experience is reported using several standard data sets found in the literature as well as using synthetically generated larger problem instances. The results validate the robustness of the proposed algorithmic procedure and exhibit its dominance over the popular k-means clustering technique. Finally, a heuristic procedure to obtain a good quality solution at a relative ease of computational effort is also described.     

The New k-Windows Algorithm for Improving thek -Means Clustering Algorithm

The process of partitioning a large set of patterns into disjoint and homogeneous clusters is fundamental in knowledge acquisition. It is called Clustering in the literature and it is applied in various fields including data mining, statistical data analysis, compression and vector quantization. The k-means is a very popular algorithm and one of the best for implementing the clustering process. The k-means has a time complexity that is dominated by the product of the number of patterns, the number of clusters, and the number of iterations. Also, it often converges to a local minimum. In this paper, we present an improvement of the k-means clustering algorithm, aiming at a better time complexity and partitioning accuracy. Our approach reduces the number of patterns that need to be examined for similarity, in each iteration, using a windowing technique. The latter is based on well known spatial data structures, namely the range tree, that allows fast range searches.     

Solving linear fractional bilevel programs

In this paper, we prove that an optimal solution to the linear fractional bilevel programming problem occurs at a boundary feasible extreme point. Hence, the Kth-best algorithm can be proposed to solve the problem. This property also applies to quasiconcave bilevel problems provided that the first level objective function is explicitly quasimonotonic.     

Automatic domain decomposition for semiclassical Bohmian mechanics using k-means clustering

We propose a method to automatically decompose domains in the context of semiclassical Bohmian mechanics. The algorithm is based on the approximate quantum potential method and the technique of k-means clustering. Two numerical examples, static analysis of quantum forces for a Pearson Type IV distribution and temporal analysis of the scattering on the Eckart barrier, are presented to show the viability of the method. The first example demonstrates that approximate quantum forces using our domain decomposition technique achieves convergence as the number of domains increases. In the second example, it is demonstrated that the domains constructed from k-means clustering has well adapted themselves to the evolving wave packet, providing coverage to both transmission and reflection waves. We also confirm that the use of multiple domains improves the evolution of the wave packet by comparing the result with the quantum mechanical solution, previously obtained. The computational cost remains manageable even with a naive implementation of time-consuming summation routines, but development of more sophisticated methodology is recommended for large scale, multidimensional calculations.     

Soliton solutions for (2 + 1)-dimensional and (3 + 1)-dimensional K(m, n) equations

We consider the nonlinear dispersive K(m,n) equation with the generalized evolution term and derive analytical expressions for some conserved quantities. By using a solitary wave ansatz in the form of sech^p function, we obtain exact bright soliton solutions for (2 + 1)-dimensional and (3 + 1)-dimensional K(m,n) equations with the generalized evolution terms. The results are then generalized to multi-dimensional K(m,n) equations in the presence of the generalized evolution term. An extended form of the K(m,n) equation with perturbation term is investigated. Exact bright soliton solution for the proposed K(m,n) equation having higher-order nonlinear term is determined. The physical parameters in the soliton solutions are obtained as function of the dependent model coefficients.     

An algebraic multigrid-based algorithm for circuit clustering

Digital circuits have grown exponentially in their sizes over the past decades. To be able to automate the design of these circuits, efficient algorithms are needed. One of the challenging stages of circuit design is the physical design where the physical locations of the components of a circuit are determined. Coarsening or clustering algorithms have become popular with physical designers due to their ability to reduce circuit sizes in the intermediate design steps such that the design can be performed faster and with higher quality. In this paper, a new clustering algorithm based on the algebraic multigrid (AMG) technique is presented. In the proposed algorithm, AMG is used to assign weights to connections between cells of a circuit and find cells that are best suited to become the initial cells for clusters, seed cells. The seed cells and the weights between them and the other cells are then used to cluster the cells of a circuit. The analysis of the proposed algorithm proves linear-time complexity, O(N), where N is the number of pins in a circuit. The numerical experiments demonstrate that AMG-based clustering can achieve high quality clusters and improve circuit placement designs with low computational cost.     

Multidimensional cluster stability analysis from a Brazilian Bradyrhizobium sp. RFLP/PCR data set

The taxonomy of the N₂-fixing bacteria belonging to the genus Bradyrhizobium is still poorly refined, mainly due to conflicting results obtained by the analysis of the phenotypic and genotypic properties. This paper presents an application of a method aiming at the identification of possible new clusters within a Brazilian collection of 119 Bradyrhizobium strains showing phenotypic characteristics of B. japonicum and B. elkanii. The stability was studied as a function of the number of restriction enzymes used in the RFLP-PCR analysis of three ribosomal regions with three restriction enzymes per region. The method proposed here uses clustering algorithms with distances calculated by average-linkage clustering. Introducing perturbations using sub-sampling techniques makes the stability analysis. The method showed efficacy in the grouping of the species B. japonicum and B. elkanii. Furthermore, two new clusters were clearly defined, indicating possible new species, and sub-clusters within each detected cluster.     

A Global Optimization RLT-based Approach for Solving the Fuzzy Clustering Problem

The field of cluster analysis is primarily concerned with the partitioning of data points into different clusters so as to optimize a certain criterion. Rapid advances in technology have made it possible to address clustering problems via optimization theory. In this paper, we present a global optimization algorithm to solve the fuzzy clustering problem, where each data point is to be assigned to (possibly) several clusters, with a membership grade assigned to each data point that reflects the likelihood of the data point belonging to that cluster. The fuzzy clustering problem is formulated as a nonlinear program, for which a tight linear programming relaxation is constructed via the Reformulation-Linearization Technique (RLT) in concert with additional valid inequalities. This construct is embedded within a specialized branch-and-bound (B&B) algorithm to solve the problem to global optimality. Computational experience is reported using several standard data sets from the literature as well as using synthetically generated larger problem instances. The results validate the robustness of the proposed algorithmic procedure and exhibit its dominance over the popular fuzzy c-means algorithmic technique and the commercial global optimizer BARON.     

Self-learning K-means clustering: a global optimization approach

An appropriate distance is an essential ingredient in various real-world learning tasks. Distance metric learning proposes to study a metric, which is capable of reflecting the data configuration much better in comparison with the commonly used methods. We offer an algorithm for simultaneous learning the Mahalanobis like distance and K-means clustering aiming to incorporate data rescaling and clustering so that the data separability grows iteratively in the rescaled space with its sequential clustering. At each step of the algorithm execution, a global optimization problem is resolved in order to minimize the cluster distortions resting upon the current cluster configuration. The obtained weight matrix can also be used as a cluster validation characteristic. Namely, closeness of such matrices learned during a sample process can indicate the clusters readiness; i.e. estimates the true number of clusters. Numerical experiments performed on synthetic and on real datasets verify the high reliability of the proposed method.     

Variable neighborhood search for harmonic means clustering

Harmonic means clustering is a variant of minimum sum of squares clustering (which is sometimes called K-means clustering), designed to alleviate the dependance of the results on the choice of the initial solution. In the harmonic means clustering problem, the sum of harmonic averages of the distances from the data points to all cluster centroids is minimized. In this paper, we propose a variable neighborhood search heuristic for solving it. This heuristic has been tested on numerous datasets from the literature. It appears that our results compare favorably with recent ones from tabu search and simulated annealing heuristics.     

On the Isomorphism Conjecture in algebraic K-theory

The Isomorphism Conjecture is a conceptional approach towards a calculation of the algebraic K-theory of a group ring RΓ, where Γ is an infinite group. In this paper we prove the conjecture in dimensions n<2 for fundamental groups of closed Riemannian manifolds with strictly negative sectional curvature and arbitrary coefficient rings R. If R is regular this leads to a concrete calculation of low dimensional K-theory groups of RΓ in terms of the K-theory of R and the homology of the group.     

Positivity of Riesz functionals and solutions of quadratic and quartic moment problems

We employ positivity of Riesz functionals to establish representing measures (or approximate representing measures) for truncated multivariate moment sequences. For a truncated moment sequence y, we show that y lies in the closure of truncated moment sequences admitting representing measures supported in a prescribed closed set K⊆Rn if and only if the associated Riesz functional Ly is K-positive. For a determining set K, we prove that if Ly is strictly K-positive, then y admits a representing measure supported in K. As a consequence, we are able to solve the truncated K-moment problem of degree k in the cases: (i) (n,k)=(2,4) and K=R²; (ii) n?1, k=2, and K is defined by one quadratic equality or inequality. In particular, these results solve the truncated moment problem in the remaining open cases of Hilbert's theorem on sums of squares.     

Using the K5Connected Cognition Diagram to analyze teachers’ communication and understanding of regions in three-dimensional space

This paper reports on a study that introduces and applies the K₅Connected Cognition Diagram as a lens to explore video data showing teachers’ interactions related to the partitioning of regions by axes in a three-dimensional geometric space. The study considers “semiotic bundles” ( Arzarello, 2006), introduces “semiotic connections,” and discusses the fundamental role each plays in developing individual understanding and communication with peers. While all teachers solved the problem posed, many failed to make or verbalize connections between the types of semiotic resources introduced during their discussions.     

Polynomial sequential continuity on C(K,E) spaces

We show that, for bounded sequences in C(K,E), the polynomial sequential convergence is not equivalent to the pointwise polynomial sequential convergence. We introduce several conditions on E under which different versions of the result are true when K is a scattered compact space. These conditions are related with some others appeared in the literature and they seem to be of independent interest.     

Strong jump-traceability I: The computably enumerable case

Recent investigations in algorithmic randomness have lead to the discovery and analysis of the fundamental class K of reals called the K-trivial reals, defined as those whose initial segment complexity is identical with that of the sequence of all 1's. There remain many important open questions concerning this class, such as whether there is a combinatorial characterization of the class and whether it coincides with possibly smaller subclasses, such as the class of reals which are not sufficiently powerful as oracles to cup a Turing incomplete Martin-Löf random real to the halting problem. Hidden here is the question of whether there exist proper natural subclasses of K. We show that the combinatorial class of computably enumerable, strongly jump-traceable reals, defined via the jump operator by Figueira, Nies and Stephan [Santiago Figueira, André Nies, Frank Stephan, Lowness properties and approximations of the jump, Electr. Notes Theor. Comput. Sci. 143 (2006) 45-57], is such a class, and show that like K, it is an ideal in the computably enumerable degrees. This is the first example of a class of reals defined by a “cost function” construction which forms a proper subclass of K. Further, we show that every c.e., strongly jump-traceable set is not Martin-Löf cuppable, thus giving a combinatorial property which implies non-ML cuppability.     

Maximal exponents of polyhedral cones (I)

Let K be a proper (i.e., closed, pointed, full convex) cone in Rn. An n×n matrix A is said to be K-primitive if there exists a positive integer k such that ; the least such k is referred to as the exponent of A and is denoted by γ(A). For a polyhedral cone K, the maximum value of γ(A), taken over all K-primitive matrices A, is called the exponent of K and is denoted by γ(K). It is proved that if K is an n-dimensional polyhedral cone with m extreme rays then for any K-primitive matrix A, γ(A)?(mA−1)(m−1)+1, where mA denotes the degree of the minimal polynomial of A, and the equality holds only if the digraph (E,P(A,K)) associated with A (as a cone-preserving map) is equal to the unique (up to isomorphism) usual digraph associated with an m×m primitive matrix whose exponent attains Wielandt's classical sharp bound. As a consequence, for any n-dimensional polyhedral cone K with m extreme rays, γ(K)?(n−1)(m−1)+1. Our work answers in the affirmative a conjecture posed by Steve Kirkland about an upper bound of γ(K) for a polyhedral cone K with a given number of extreme rays.