Doutorado em Engenharia Elétrica

URI Permanente para esta coleção

http://repositorio.ufes.br/handle/10/17618

Nível: Doutorado
Ano de início:
Conceito atual na CAPES:
Ato normativo:
Periodicidade de seleção:
Área(s) de concentração:
Url do curso:

Navegar

Navegando Doutorado em Engenharia Elétrica por Autor "Almeida, Ailson Rosetti de"

Agora exibindo 1 - 2 de 2
  • Nenhuma Miniatura disponível
    Item
    Implementação de métodos biométricos bi-modais baseados em fusão de características para reconhecimento de indivíduos
    (Universidade Federal do Espírito Santo, 2007-08-24) Salomão, João Marques; Salles, Evandro Ottoni Teatini; Almeida, Ailson Rosetti de; Baptista, Edson; Montalvão Filho, Jugurta Rosa; Sarcinelli Filho, Mário; Bastos Filho, Teodiano Freire
    This research proposes a bi-modal personal recognition system based on the information fusion in the level of the characteristics (also called features). The biometric and behavioral features considered in the validation of our proposal are obtained from the video sequences of the human gait and face images. It evaluates some proposals of architectures and algorithms in the implementation of multiple biometric methods for recognition and authentication of individuals through the fusion of face features and the human gait, whose research subject is scarcely available in the scientific literature. The choice of the approach for the fusion of the two particularly considered information sources (not to be confused with multiple attributes of a feature vector from one information source) is justified by the supposition that it must offer a better classification performance, robustness and safety as it allows the night identification at a certain distance. Furthermore, it is considered less intrusive than all the other biometric systems since it presents little or no need for the collaboration of a person to be identified. The research, being an state of the art theme, has few bibliographical references and starts by evaluating the performance of the human recognition system through gait as it applies, on the video sequences, techniques of extraction and selection of silhouette features based on PCA (Principal Component Analysis), ICA (Independent Component Analysis), Wavelet transforms and PoV (Proportion of Variances), combined with the classifiers based on Euclidian distances, SVM (Support Vector Machines) and neural networks with RBF (Radial Base Functions). Using databases of public domain, a special study was carried out on gait aiming at an initial evaluation of its performance in human recognition systems, as well as the choice of methods and techniques more appropriate to make the fusion of this gait with the face images of the individual. As an improvement, on a second step, own databases were obtained based on an experimental setup that allowed the extraction of silhouettes from gait video sequences as well as face images corresponding to each sequence. This research also evaluated the performance of the recognition system when determining the fusion of both biometric features. Then, after the definition of the most adequate architecture, the fusion of two biometric systems face-gait on public domain databases, as well as our particular one, is proposed and implemented. In this final phase of the work, the proposed implementations use extraction techniques and features selection based on gait silhouettes energy, in the proportion of variances (PoV) and in the classification algorithms based on neural networks with radial base functions (RBF). The obtained results allowed the evaluation of the non-error rates, the false acceptance (FAR) and rejection rates (FRR) and the system performance, when the vectors of individual features of the face images alone, the gait video sequences alone, as well as the fusion of both are considered, confirming the validity and viability of the application of our proposal of fusion in the feature level in the bimodal human recognition system
  • Nenhuma Miniatura disponível
    Item
    Modelo de Seleção de Carteiras Baseado em Erros de Predição
    (Universidade Federal do Espírito Santo, 2008-12-18) Freitas, Fábio Daros de; Souza, Alberto Ferreira de; Almeida, Ailson Rosetti de; Salles, Evandro Ottoni Teatini; Zandonade, Eliana; França, Felipe Maia Galvão
    This work presents a new prediction errors-based portfoliooptimization model that cap-tures short-term investment opportunities. We used autoregressive moving references neuralnetwork predictors to predict the stock’s returns and derived a risk measure based on thepredictor’s errors of prediction that maintains the same statistical foundation of the mean-variance model. The efficient diversification effects hold by selecting predictors with lowand complimentary error profiles. A large set of experimentswith real data from the Brazil-ian stock market was employed to evaluate our portfolio optimization model, which includedthe examination of the Normality of the errors of prediction. Our main results showed that itis possible to obtain Normal prediction errors with non-Normal series of stock returns, andthat the prediction errors-based portfolio optimization model better captured the short termopportunities, outperforming the mean-variance model andbeating the market index.