Academic
Publications
X-means: Extending K-means with Efficient Estimation of the Number of Clusters

X-means: Extending K-means with Efficient Estimation of the Number of Clusters,Dan Pelleg,Andrew W. Moore

X-means: Extending K-means with Efficient Estimation of the Number of Clusters   (Citations: 403)
BibTex | RIS | RefWorks Download
Despite its popularity for general clustering,K-means suffers three major shortcomings;it scales poorly computationally, the numberof clusters K has to be supplied by theuser, and the search is prone to local minima.We propose solutions for the first twoproblems, and a partial remedy for the third.Building on prior work for algorithmic accelerationthat is not based on approximation,we introduce a new algorithm that efficiently,searches the space of cluster locations and...
Cumulative Annual
    • ...Pelleg and Moore [11] proposed an approach to find the best partitioning of the data where the average variance of the clusters is minimum...
    • ... partitionings ( x 1 ). To evaluate these partitionings, we compute the BIC score of each partitioning, the highest BIC means best available partitioning of the execution trace and consequently the best estimation of the number of clusters K, and which also corresponds to the number of identified phases Since the dimension of the data in our case is 1, we use a special formulation of the BIC (for a more general case of BIC formulation see ...
    • ...log-likelihood of the data according to the j-th partitioning which can be computed as follows (see [11] for more details on using BIC formulation in k-means clustering):...

    Heidar Pirzadehet al. A Novel Approach Based on Gestalt Psychology for Abstracting the Conte...

    • ...The X-Means[4] clustering algorithm was used to group all similar weeks and extract a number of outliers by using these data signatures...

    Jasmine A. Malinaoet al. A Quantitative Analysis-Based Algorithm for Optimal Data Signature Con...

    • ...In order to find common patterns between these instances, we carried out a k-means clustering based on an improved extension of the basic k-means algorithm, developed by Pelleg and Moore (2000) and implemented in the Weka platform (Hall et al, 2009)...

    Luiz Navedaet al. Microtiming Patterns and Interactions with Musical Properties in Samba...

    • ... random number of points is chosen and a fixed-scale and -orientation variant of the SIFT descriptors [19] is usedtocharacterize them.Theglobalset ofdescriptorsisthen clustered using k-means (see, e.g., [20]) with .T his value was set after an initial round of experiments as the best foundgiven thenumber of objects,sequencesand characterstics of the dB. (Notice that the optimal value of could be found automatically, e.g., using x-means [ ...

    C. Castelliniet al. Using Object Affordances to Improve Object Recognition

    • ...Regarding the number of clusters in the k-means, Pelleg and Moore developed the X-means algorithm to find the number of clusters existing in the data set automatically, using the Bayesian Information Criterion (BIC) to evaluate the cluster distribution [12]...

    Aislan G. Foinaet al. P-means, a parallel clustering algorithm for a heterogeneous multi-pro...

Sort by: