Text Categorisation Through Dimensionality Reduction Using Wavelet Transform

This paper proposes a new method of dimensionality reduction when performing Text Classification, by applying the discrete wavelet transform to the document-term frequencies matrix. We analyse the features provided by the wavelet coefficients from the different orientations: (1) The high energy coefficients in the horizontal orientation correspond to relevant terms in a single document. (2) The high energy coefficients in the vertical orientation correspond to relevant terms for a single document, but not for the others. (3) The high energy coefficients in the diagonal orientation correspond to relevant terms in a document in comparison to other terms. If we filter using the wavelet coefficients and fulfil these three conditions simultaneously, we can obtain a reduced vocabulary of the corpus, with less dimensions than in the original one. To test the success of the reduced vocabulary, we recoded the corpus with the new reduced vocabulary and we obtained a statistically relevant level of accuracy for document classification.

FROM WINKLER’S FOUNDATION TO POPOV’S FOUNDATION

In recent years, the method of dimensionality reduction (MDR) has started to figure as a very convenient tool for dealing with a wide class of elastic contact problems. The MDR modeling framework introduces an equivalent punch profile and a one-dimensional Winkler-type elastic foundation, called henceforth Popov’s foundation. While the former mainly accounts for the geometry of contact configuration, the Popov foundation inherits the main characteristics of both the contact interface (like friction and adhesion) and the contacting elastic bodies (e.g., anisotropy, viscoelasticity or inhomogeneity). The discussion is illustrated with an example of the Kendall-type adhesive contact for an isotropic elastic half-space.

NUMERICAL IMPLEMENTATION OF FRETTING WEAR IN THE FRAMEWORK OF THE MDR

Two numerical methods are proposed to improve accuracy of the numerical calculation of fretting wear in the framework of the Method of Dimensionality Reduction (MDR). Due to the singularity of the transformation equations, instabilities appear at the border between the stick and slip regions after many transformations from the one-dimensional to the three-dimensional contact and back. In these two methods, the transformation equations are reformulated to weaken the singularity of the integrals and a stable simulation of fretting wear is realized even with the wear models which go beyond the classical Archard law. With an example of dual-oscillation, we show the change in the worn profile of a parabolic indenter as well as the stress distribution on the contacting surface during the oscillating cycles under the Archard’s law of wear and Coulomb’s law of friction.

NUMERICAL METHODS FOR THE SIMULATION OF DEFORMATIONS AND STRESSES IN TURBINE BLADE FIR-TREE CONNECTIONS

In this work, different numerical methods for simulating deformations and stresses in turbine blade fir-tree connections are examined. The main focus is on the Method of Dimensionality Reduction (MDR) and the Boundary Element Method (BEM). Generally, the fir-tree connections require a computationally expensive finite element setup. Their complex geometry exceeds the limitations of the faster numerical techniques which are used with great success within the framework of the half-space approximation. Ways of extending the application range of the MDR and the BEM to the particular problem of highly undulating surfaces of the fir-tree connection are shown and discussed.

METHOD OF DIMENSIONALITY REDUCTION IN CONTACT MECHANICS AND FRICTION: A USER’S HANDBOOK. III. VISCOELASTIC CONTACTS

Until recently the analysis of contacts in tribological systems usually required the solution of complicated boundary value problems of three-dimensional elasticity and was thus mathematically and numerically costly. With the development of the so-called Method of Dimensionality Reduction (MDR) large groups of contact problems have been, by sets of specific rules, exactly led back to the elementary systems whose study requires only simple algebraic operations and elementary calculus. The mapping rules for axisymmetric contact problems of elastic bodies have been presented and illustrated in the previously published parts of The User's Manual, I and II, in Facta Universitatis series Mechanical Engineering [5, 9]. The present paper is dedicated to axisymmetric contacts of viscoelastic materials. All the mapping rules of the method are given and illustrated by examples.

A method of dimensionality reduction by selection of components in principal component analysis for text classification

Dimensionality reduction, including feature extraction and selection, is one of the key points for text classification. In this paper, we propose a mixed method of dimensionality reduction constructed by principal components analysis and the selection of components. Principal components analysis is a method of feature extraction. Not all of the components in principal component analysis contribute to classification, because PCA objective is not a form of discriminant analysis (see, e.g. Jolliffe, 2002). In this context, we present a function of components selection, which returns the useful components for classification by the indicators of the performances on the different subsets of the components. Compared to traditional methods of feature selection, SVM classifiers trained on selected components show improved classification performance and a reduction in computational overhead.