Clustering suffers from the curse of dimensionality, and similarity functions that use all input features with equal relevance may not be effective. We introduce an algorithm that discovers clusters in subspaces spanned by different combinations of dimensions via local weightings of features. This approach avoids the risk of loss of information encountered in global dimensionality reduction tec...

We present a novel algorithm for automatically co-segmenting a set of shapes from a common family into consistent parts. Starting from over-segmentations of shapes, our approach generates the segmentations by grouping the primitive patches of the shapes directly and obtains their correspondences simultaneously. The core of the algorithm is to compute an affinity matrix where each entry encodes ...

We propose a transformation method to circumvent the problems with high dimensional data. For each object in the data, we create an itemset of the k-nearest neighbors of that object, not just for one of the dimensions, but for many views of the data. On the resulting collection of sets, we can mine frequent itemsets; that is, sets of points that are frequently seen together in some of the views...

VALCRI provides a challenging and overwhelming high-dimensional dataset that comprises of hundreds of extracted semantic features in addition to the usual spatiotemporal information or metadata. To overcome the curse of dimensionality and to generate low-dimensional representations of these semantic features we apply interactive high-dimensional data analysis techniques with the goal of obtaini...

Résumé. Cet article se place dans le cadre du subspace clustering, dont la problématique est double : identifier simultanément les clusters et le sousespace spécifique dans lequel chacun est défini, et caractériser chaque cluster par un nombre minimal de dimensions, permettant ainsi une présentation des résultats compréhensible par un expert du domaine d’application. Les méthodes proposées jusq...

Traditional similarity measurements often become meaningless when dimensions of datasets increase. Subspace clustering has been proposed to find clusters embedded in subspaces of high dimensional datasets. Many existing algorithms use a grid based approach to partition the data space into nonoverlapping rectangle cells, and then identify connected dense cells as clusters. The rigid boundaries o...

Dimensionality reduction methods are very common in the field of high dimensional data analysis, where the classical analysis methods are inadequate. Typically, algorithms for dimensionality reduction are computationally expensive. Therefore, their applications to process data warehouses are impractical. It is visible even more when the data is accumulated non-stop. In this paper, an out-of-sam...

Appendix A contains proofs for our main results. The proofs are sorted in the order that their corresponding statements appear in the paper. Appendix B formalizes our claims in the paper about attribute privacy and the corresponding utility theorem and includes additional discussions on the difficulty of a stronger user-level privacy claim. Appendix C contains numerical simulations on the perfo...