Computing evolutionary chains in musical sequences.
The paper briefly reviews recent advances in the methodology of feature selection (FS) and the conceptual base of a consulting system for solving FS problems. The reasons for designing a kind of expert or consulting system which would guide a less experienced user are outlined. The paper also attempts to provide a guideline which approach to choose with respect to the extent of a priori knowledge of the problem. The methods discussed here form the core of the software package being developed for...
In classification problems, the issue of high dimensionality, of data is often considered important. To lower data dimensionality, feature selection methods are often employed. To select a set of features that will span a representation space that is as good as possible for the classification task, one must take into consideration possible interdependencies between the features. As a trade-off between the complexity of the selection process and the quality of the selected feature set, a pairwise...
La représentation visuelle d'une hiérarchie induit un ordre sur les singletons. Si l'on désire représenter la même hiérarchie en tenant compte de contraintes extérieures (ordre des singletons induit par une autre hiérarchie, une partition, un indice de dissimilarité, par exemple) des croisements peuvent apparaître. Il y a un croisement dans la représentation visuelle d'une hiérarchie quand une branche horizontale (associée à un palier) est coupée par une branche verticale associée à un singleton....
In this paper we present a novel approach to decomposing high dimensional spaces using a multiobjective genetic algorithm for identifying (near-)optimal subspaces for hierarchical classification. This strategy of pre-processing the data and explicitly optimising the partitions for subsequent mapping onto a hierarchical classifier is found to both reduce the learning complexity and the classification time with no degradation in overall classification error rate. Results of partitioning pattern spaces...
Detecting a cluster structure is considered. This means solving either the problem of discovering a natural decomposition of data points into groups (clusters) or the problem of detecting clouds of data points of a specific form. In this paper both these problems are considered. To discover a cluster structure of a specific arrangement or a cloud of data of a specific form a class of nonlinear projections is introduced. Fitness functions that estimate to what extent a given subset of data points...
searching for all occurrences of a pattern (string or sequence) in some text, where the pattern can occur with some limited number of errors given by edit distance. Several methods were designed for the approximate string matching that simulate nondeterministic finite automata (NFA) constructed for this problem. This paper presents reduced NFAs for the approximate string matching usable in case, when we are interested only in occurrences having edit distance less than or equal to a given integer,...
We present a novel eye localization method which can be used in face recognition applications. It is based on two SVM classifiers which localize the eyes at different resolution levels exploiting the Haar wavelet representation of the images. We present an extensive analysis of its performance on images of very different public databases, showing very good results.
In this paper the possibilities are discussed for training statistical pattern recognizers based on a distance representation of the objects instead of a feature representation. Distances or similarities are used between the unknown objects to be classified with a selected subset of the training objects (the support objects). These distances are combined into linear or nonlinear classifiers. In this approach the feature definition problem is replaced by finding good similarity measures. The proposal...