http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Qiu-yue Tai,Kyung-shik Shin(신경식) 한국지능정보시스템학회 2010 한국지능정보시스템학회 학술대회논문집 Vol.2010 No.11월
The back-propagation neural network (BPN) has been broadly applied to financial distress prediction because their excellent treatment of nonlinear data with learning capabilities. Despite the wide application of BPN, some major issues must be considered before its use, such as the network topologies, learning parameters, and normalization methods for the input and output vectors. Previous studies on bankruptcy prediction with BPN have shown, however, that many researchers are interested in how to optimize the network topologies and learning parameters to improve the network’s prediction performance. In many cases, the benefits of data normalization are overlooked. The most representative method of normalization for BPN is linear scaling, which can reduce the dimensionality of the input space, thus helping speed up the learning phase and improve the classification performance. This method, however, has the limitation of only adjusting the scale of the original data and of not being able to relieve the complicated relationships among the data. An alternative method involves applying the fuzzy set theory to normalize the data for the neural network, because the fuzzy membership function can represent the continuous and complicated values as degree of membership values, and allows the representation of the concepts that can be regarded as falling under more than one category. In this study, a genetic algorithm (GA)-optimized nonlinear fuzzy normalization method was proposed. The polynomial-based nonlinear fuzzy membership function was used to normalize the data within the value of [0, 1]. GA was thus used to find the optimal boundary value of the fuzzy parameters. Based on the results of the experiment that was conducted, the proposed method was evaluated and compared with other methods to demonstrate its advantage.
Qiu-yue Tai(태추월),Kyung-shik Shin(신경식) 한국지능정보시스템학회 2010 지능정보연구 Vol.16 No.3
The back-propagation neural network (BPN) has long been successfully applied in bankruptcy prediction problems. Despite its wide application, some major issues must be considered before its use, such as the network topology, learning parameters and normalization methods for the input and output vectors. Previous studies on bankruptcy prediction with BPN have shown that many researchers are interested in how to optimize the network topology and learning parameters to improve the prediction performance. In many cases, however, the benefits of data normalization are often overlooked. In this study, a genetic algorithm (GA)-based normalization transform, which is defined as a linearly weighted combination of several different normalization transforms, will be proposed. GA is used to extract the optimal weight for the generalization. From the results of an experiment, the proposed method was evaluated and compared with other methods to demonstrate the advantage of the proposed method.