This paper considers distributed clustering of high dimensional heterogeneous data using a distributed Principal Component Analysis (PCA) technique called the Collective PCA. It presents the Collective PCA technique that can be used independent of the clustering application. It shows a way to integrate the Collective PCA with a given o-the-shelf clustering algorithm in order to develop a distributed clustering technique. It also presents experimental results using dierent test data sets including an application for web mining.

This report is a masters thesis written at the Department of Mathematics, Linköping University. Two different classification algorithms for handwritten digit recognition have been thoroughly analysed. The first algorithm uses Higher Order Singular Value Decomposition (HOSVD) of the training digits. The second algorithm relies on a specific distance measure, which is invariant to different transformations, called Tangent Distance (TD). This algorithm was modified in the implementation part by the use of numerical derivatives and an approximation of the blurring operator. Two more classification algorithms were constructed by combining the first two algorithms. All constructed algorithms have been tested with good performance for some of them. The best results were achieved by the Tangent Distance classifier with an error rate of 3 %. Finally the results of a few other classifiers are presented and compared with the test results obtained in this report.

In this paper we present two algorithms for handwritten digit classification based on the higher order singular value decomposition (HOSVD). The first algorithm uses HOSVD for construction of the class models and achieves classification results with error rate lower than 6%. The second algorithm uses the HOSVD for tensor approximation simultaneously in two modes. Classification results for the second algorithm are almost down at 5 % even though the approximation reduces the original training data with more than 98 % before the construction of the class models. The actual classification in the test phase for both algorithms is conducted by solving a series least squares problems. Considering computational amount for the test presented the second algorithm is twice as efficient as the first one.

This work considers the application of several related multivariate data analysis techniques to the monitoring and modeling of dynamic processes. Included are the method of Principal Components Analysis (PCA), and the regression technique Continuum Regression (CR), which encompasses Principal Components Regression (PCR), Partial Least Squares (PLS) and Multiple Linear Regression (MLR), all of which are based on eigenvector decompositions. It is shown that proper application of PCA to the measurements from multivariate processes can facilitate the detection of failed sensors and process upsets. The relationship between PCA and the state-space process model form is shown, providing a theoretical basis for the use of PCA in dynamic systems. For processes with more measurements than sta...

This Provisional PDF corresponds to the article as it appeared upon acceptance. Fully formatted PDF and full text (HTML) versions will be made available soon. Piecewise multivariate modelling of sequential metabolic profiling data BMC Bioinformatics 2008, 9:105 doi:10.1186/1471-2105-9-105

Abstract not found

In this paper we propose a robust classification rule for skewed distributions. For low dimensional data, the classification is based on the adjusted outlyingness. In the case of high dimensional data, the robustified SIMCA method is adjusted for skewness. The robustness of the method is investigated through different simulations and by applying it to various real examples.

This paper outlines our submissions to different music classification tasks for the Music Information Retrieval Evaluation eXchange (MIREX) 2009. We detail here three different algorithms tested in mood and genre classification tasks, and in classical composer identification. These algorithms

Audio content processing for automatic music genre classification: descriptors,

Abstract. A key factor in classifiers based on the normal (or Gaussian) distribution is the modeling of covariance matrices. When the number of available training pixels is limited, as often is the case in hyperspectral image classification, it is necessary to limit the complexity of these covariance models. An alternative to reducing the complexity uniformly over the whole feature space, is to form orthogonal subspaces and reduce the model complexity within them separately, e.g., forming full-complexity within-class, or interior-class, subspace models, and reduced-complexity exterior-class subspace models. We propose to use subspaces created by forming fewer and wider spectral bands, instead of the more general principal component analysis transform (PCA), in an attempt to exploit a-priori knowledge of the data to create more generalizable subspaces. We investigate the resulting classifiers by studying their performances on four hyperspectral data sets. On each data set, experiments where run using different training set sizes. The results indicate that the classifiers seem to benefit from using this more data-specific approach to forming subspaces. 1