Time filter

Source Type

Zhou J.-D.,Air Force Equipment Software Testing Center
Tien Tzu Hsueh Pao/Acta Electronica Sinica

To balance the diversity and the accuracy in ensemble learning and improve the generalization performance of learning system, a selective ensemble algorithm based on Bagging and confusion matrix is proposed. By disturbing the training set and feature space, base classifiers are generated, of which the confusion matrix is used to construct a measure matrix of the diversity between classifiers; then based on the measure matrix, base classifier set is divided into different subsets, from which each base classifier is selected for ensemble; finally, majority voting method is utilized to fusion the base classifiers' recognition results and experiments have been done to attest the validity of the proposed algorithm. Source

Lei L.,PLA Air Force Aviation University | Wang X.-D.,PLA Air Force Aviation University | Luo X.,PLA Air Force Aviation University | Zhou J.-D.,Air Force Equipment Software Testing Center
Tien Tzu Hsueh Pao/Acta Electronica Sinica

Multi-classification has been one of the research hotspot in pattern recognition, and there are many solutions to it. As a common way to model multi-classification to design a set of binary classifiers and fuse them, Error-correcting output codes(ECOC) represents a successful framework to deal with this type of problems and is attracting more and more attention of researchers. In this paper, the framework of ECOC is concluded at first. Then the two keys of multi-classification based on ECOC, i. e., the coding strategies and decoding strategies are proposed. The main part focuses on the research of the two keys and the application of ECOC. Finally, the still existing problems of ECOC are pointed out and the promising research fields are given. The analysis of the paper will provide reference and advice in the practical application of multi-classification based on ECOC. ©, 2014, Chinese Institute of Electronics. All right reserved. Source

Zhou J.-D.,Air Force Equipment Software Testing Center | Zhou H.-J.,Air Force Equipment Software Testing Center | Yang Y.,Air Force Equipment Software Testing Center | Hu H.-Y.,Military Representatives Office of PLA
Tien Tzu Hsueh Pao/Acta Electronica Sinica

It is known that error-correcting output codes (ECOC) is a common way to model multiclass classification problems, in which the research of encoding based on data especially attracts attentions. In this paper, we proposed a method for learning error-correcting output codes with the help of a single layered perception neural network. To achieve this goal, the code elements of ECOC are mapped to the weights of network for the given decoding strategy, and an object function with the constrained weights used as a cost function of network. After the training, we can obtain a coding matrix including lots of subgroups of class. Experimental results on artificial data and UCI with logistic linear classifier (LOGLC) as the binary learner show that our scheme provides better performance of classification with shorter length of coding matrix than other state-of-the-art encoding strategies. Source

Discover hidden collaborations