Failure prediction of banks using threshold accepting trained kernel principal component neural network

Differential evolution trained kernel principal component WNN and kernel binary quantile regression: Application to banking

“…Thus, we have 11 predictions of the response variable [64], PCA-TAWNN [76] and DEWNN [17]. For the US Bankruptcy dataset, we compare the results with that of DEWNN only.…”

“…They employed PCNN for bankruptcy prediction problems and reported that PCNN outperformed BPNN, Threshold Accepting trained Neural Network (TANN), PCA-BPNN and PCA-TANN in terms of area under receiver operating characteristic curve (AUC) criterion. Later, to solve bankruptcy prediction problems, Ravisankar and Ravi [64] employed KPCNN trained by threshold accepting based training algorithm. KPCNN is a nonlinear version of the PCNN.…”

“…Ravisankar and Ravi [64] proposed KPCNN classifier trained by threshold accepting with different kernels such as polynomial, sigmoid and Gaussian. They successfully applied this architecture to predict bankruptcy in banks.…”

“…Li et al [42] proposed Kernel Quantile Regression (KQR) in which he employed Quantile Regression (QR) in reproducing kernel Hilbert spaces. Ravisankar and Ravi [64] proposed a new architecture called Kernel Principal Component Neural Network fed as input to BQR. The effectiveness of the proposed techniques is tested on several benchmark datasets and bank datasets.…”

Financial crisis predictions based on biorthogonal wavelet hybrid kernel support vector machine

“…Therefore, kernel principal component analysis (KPCA) has been adopted to make feature subset selection of financial indexes, and a satisfying result has been got [8] [9]. Since that KPCA conducts feature extraction by mapping data to high-dimensional feature space through kernel functions, the selection of kernel functions will directly affect the feature extraction of financial indexes.…”