4 resultados para feature representation

em Repositório Institucional UNESP - Universidade Estadual Paulista "Julio de Mesquita Filho"


Relevância:

60.00% 60.00%

Publicador:

Resumo:

The merit of the Karhunen-Loève transform is well known. Since its basis is the eigenvector set of the covariance matrix, a statistical, not functional, representation of the variance in pattern ensembles is generated. By using the Karhunen-Loève transform coefficients as a natural feature representation of a character image, the eigenvector set can be regarded as an feature extractor for a classifier.

Relevância:

30.00% 30.00%

Publicador:

Resumo:

Most face recognition approaches require a prior training where a given distribution of faces is assumed to further predict the identity of test faces. Such an approach may experience difficulty in identifying faces belonging to distributions different from the one provided during the training. A face recognition technique that performs well regardless of training is, therefore, interesting to consider as a basis of more sophisticated methods. In this work, the Census Transform is applied to describe the faces. Based on a scanning window which extracts local histograms of Census Features, we present a method that directly matches face samples. With this simple technique, 97.2% of the faces in the FERET fa/fb test were correctly recognized. Despite being an easy test set, we have found no other approaches in literature regarding straight comparisons of faces with such a performance. Also, a window for further improvement is presented. Among other techniques, we demonstrate how the use of SVMs over the Census Histogram representation can increase the recognition performance.

Relevância:

30.00% 30.00%

Publicador:

Resumo:

This article seeks to reflect on geographic representation in the coats of arms of countries in Latin America, showing how the physical, aspects of the landscape, the elements of the economy and the republican symbols were used by local elites to compose an imaginary nation in the nineteenth century. This process of "naturalization of territory" was used as an important feature in the national discourse, because this time, in most cases, the Latin American nations were composed of multi-ethnic states, with strong differences of class and a large illiterate population plus a very tenuous territory from the point of view of national integration. Thus, the elements related to geographic image through the use of coats of arms, conveyed strong messages to citizens, showing how these heraldic symbols can become an important source of research to unravel the process of building the imaginary nation.

Relevância:

30.00% 30.00%

Publicador:

Resumo:

Feature selection aims to find the most important information to save computational efforts and data storage. We formulated this task as a combinatorial optimization problem since the exponential growth of possible solutions makes an exhaustive search infeasible. In this work, we propose a new nature-inspired feature selection technique based on bats behavior, namely, binary bat algorithm The wrapper approach combines the power of exploration of the bats together with the speed of the optimum-path forest classifier to find a better data representation. Experiments in public datasets have shown that the proposed technique can indeed improve the effectiveness of the optimum-path forest and outperform some well-known swarm-based techniques. © 2013 Copyright © 2013 Elsevier Inc. All rights reserved.