In this paper we introduce new algorithms for unsupervised learning based on the use of a kernel matrix. All the information required by such algorithms is contained in the eigenvectors of the matrix or of closely related matrices. We use two different but related cost functions, the Alignment and the 'cut cost'. The first one is discussed in a companion paper [3], the second one is based on gr...

We take apart, combine and compare on real and artificial data the features of the four best-known spectral clustering algorithms. We find that the algorithms behave more similarly then expected, especially if the data are near a case called perfect, where three of the algorithms are equivalent.

This paper focuses on service clustering and uses service descriptions to construct probabilistic models for service clustering. We discuss how service descriptions can be enriched with machine-interpretable semantics and then we investigate how these service descriptions can be grouped in clusters in order to make discovery, ranking, and recommendation faster and more effective. We propose usi...

Many data mining algorithms require as a pre-processing step the discretization of real-valued data. In this paper we review some discretization methods based on clustering. We describe in detail the algorithms of discretization of a continuos real-valued attribute using the hierarchical graph clustering methods.

Abstra t We investigate the use of biased sampling a ording to the density of the dataset, to speed up the operation of general data mining tasks, su h as lustering and outlier dete tion in large multidimensional datasets. In density-biased sampling, the probability that a given point will be in luded in the sample depends on the lo al density of the dataset. We propose a general te hnique for ...

Networks are used in many scientific fields such as biology, social science, and information technology. They aim at modelling, with edges, the way objects of interest, represented by vertices, are related to each other. Looking for clusters of vertices, also called communities or modules, has appeared to be a powerful approach for capturing the underlying structure of a network. In this contex...

At the heart of unsupervised clustering and semi-supervised clustering is the calculation of matrix eigenvalues(eigenvectors) or matrix inversion. In generally, its complexity is O(N). By using Krylov Subspace Methods and Fast Methods, we improve the performance to O(NlogN). We also make a thorough evaluation of errors introduced by the fast algorithm.

The clustering problem is to assign labels to points in order to group them in a structurally meaningful way. This is often accomplished by defining cluster centroids in the vector space and assigning points to the cluster with the nearest centroid, as the k-means algorithm does. But this approach does not work for clusters that have unusual shapes, particularly if the clusters are interwoven o...

Nowadays, a vast amount of spatio-temporal data are being generated by devices like cell phones, GPS and remote sensing devices and therefore discovering interesting patterns in such data became an interesting topics for researchers. One of these topics has been spatio-temporal clustering which is a novel sub field of data mining and Recent researches in this area has focused on new methods and...