Antonio Padua Braga

Antonio Padua Braga

Auto-generated profile, based on publicly available information

Antonio Padua Braga

Universidade Federal de Minas Gerais

Research Focus & Metrics

Academic Career Length

27Years (2000-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
22 February 2013 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
CCCristiano L. Castro
Cristiano L. Castro
ABAntônio P. Braga
Antônio P. Braga
Traditional learning algorithms applied to complex and highly imbalanced training sets may not give satisfactory results when distinguishing between examples of the classes. The tendency is to yield classification models that are biased towards the overrepresented (majority) class. This paper investigates this class imbalance problem in the context of multilayer perceptron (MLP) neural networks. The consequences of the equal cost (loss) assumption on imbalanced data are formally discussed from a statistical learning theory point of view. A new cost-sensitive algorithm (CSMLP) is presented to improve the discrimination ability of (two-class) MLPs. The CSMLP formulation is based on a joint objective function that uses a single cost parameter to distinguish the importance of class errors. The learning rule extends the Levenberg-Marquadt's rule, ensuring the computational efficiency of the algorithm. In addition, it is theoretically demonstrated that the incorporation of prior information via the cost parameter may lead to balanced decision boundaries in the feature space. Based on the statistical analysis of results on real data, our approach shows a significant improvement of the area under the receiver operating characteristic curve and G -mean measures of regular MLPs.

16 March 2009 in IEEE Transactions on Power Delivery
Verified
Journal Article
Closed access
CACarlos A. Laurentys Almeida
Carlos A. Laurentys Almeida
ABAntônio P. Braga
Antônio P. Braga
SNSinval Nascimento
Sinval Nascimento
VPVinicius Paiva
Vinicius Paiva
HMHélvio J. A. Martins
Hélvio J. A. Martins
WCWalmir M. Caminhas
Walmir M. Caminhas
This paper describes a methodology that aims to extract information to enable the detection and diagnosis of faults in surge arresters, using the thermovision technique. Thermovision is a non-destructive technique used in diverse services of maintenance, having the advantage not to demand the disconnection of the equipment. The methodology uses a digital image processing algorithm based on the Watershed Transform to get the segmentation of the surge arresters. By applying the methodology is possible to classify surge arresters operative condition in: faulty, normal, light, and suspicious. The computational system generated train its neuro-fuzzy network by using a historical thermovision data. During the train phase, a heuristic is proposed in order to set the number of networks in the diagnosis system. This system was validated by a database with a hundreds of different faulty sceneries. The validation error of the set of neuro-fuzzy and the automatic digital thermovision image processing was about 10%t. The diagnosis system described has been successfully used by Electric Energy Research Center as a decision making tool for surge arresters fault diagnosis.

1 November 2000 in Neurocomputing
Verified
Journal Article
Closed access
RTRoselito de Albuquerque Teixeira
Roselito de Albuquerque Teixeira
ABAntônio Pádua Braga
Antônio Pádua Braga
RTRicardo Takahashi
Ricardo Takahashi
RSRodney R. Saldanha
Rodney R. Saldanha

7 November 2002
Verified
Conference Paper
Closed access
MAM. Barros de Almeida
M. Barros de Almeida
ABA. de Padua Braga
A. de Padua Braga
JBJ.P. Braga
J.P. Braga
A procedure called SVM-KM, based on clustering by k-means and to accelerate the training of Support Vector Machine, is the main objective of the present work. During the Support Vector Machines (SVMs) optimization phase, training vectors near the separation margins are likely to become support vector and must be preserved. Conversely, training vectors far from the margins are not in general taken into account for SVM's design process. SVM-KM groups the training vector in many clusters. Clusters formed only by a vector that belongs to the same class label can be disregard and only cluster centers are used. On the other hand, clusters with more than one class label are unchanged and all training vectors belonging to them are considered. Cluster with mixed composition are likely to happen near the separation margins and they may hold some support vectors. Consequently, the number of vectors in a SVM training is smaller and the training time can be decreased without compromising the generalization capability of the SVM.

1 March 2007 in Neurocomputing
Verified
Journal Article
Closed access
MCMarcelo Azevedo Costa
Marcelo Azevedo Costa
ABAntônio de Pádua Braga
Antônio de Pádua Braga
BMBenjamin Rodrigues de Menezes
Benjamin Rodrigues de Menezes

1 November 2010 in Mechanical Systems and Signal Processing
Verified
Journal Article
Closed access
LALuis A. Aguirre
Luis A. Aguirre
BBBruno Barbosa
Bruno Barbosa
ABAntônio P. Braga
Antônio P. Braga