应用中的各种因素可能造成数据缺失,影响后续任务的分析。因此,数据集缺失值的插补尤为重要。相比原本没有插补的处理,错误的插补值也会对分析造成更严重的偏差。针对这种情况,提出新的采用双重判别器的基于条件生成对抗插补网络(C-GAIN...应用中的各种因素可能造成数据缺失,影响后续任务的分析。因此,数据集缺失值的插补尤为重要。相比原本没有插补的处理,错误的插补值也会对分析造成更严重的偏差。针对这种情况,提出新的采用双重判别器的基于条件生成对抗插补网络(C-GAIN)的缺失值插补算法DDC-GAIN(Dual Discriminator based on C-GAIN)。该算法通过一个辅助判别器辅助主判别器判断预测值的真假,即根据一个样本的全局信息判断这个样本生成的真假,更注重特征之间的关系,以此估算预测值。在4个数据集上与5种经典插补算法进行对比实验,结果表明:同样条件下,DDC-GAIN算法在样本量较大时的均方根误差(RMSE)最低;在Default credit card数据集上缺失率为15%时,DDC-GAIN算法的RMSE比次优算法C-GAIN降低了28.99%。这说明利用辅助判别器帮助主判别器学习特征之间的关系是有效的。展开更多
The new orthogonal relationship is generalized for orthotropic elasticity of three-dimensions. The thought of how dual vectors are constructed in a new orthogonal relationship for theory of elasticity is generalized i...The new orthogonal relationship is generalized for orthotropic elasticity of three-dimensions. The thought of how dual vectors are constructed in a new orthogonal relationship for theory of elasticity is generalized into orthotropic problems. A new dual vector is presented by the dual vector of the symplectic systematic methodology for elasticity that is over again sorted. A dual differential equation is directly obtained by using a mixed variables method. A dual differential matrix to be derived possesses a peculiarity of which principal diagonal sub-matrixes are zero matrixes. As a result of the peculiarity of the dual differential matrix, two independently and symmetrically orthogonal sub-relationships are discovered for orthotropic elasticity of three-dimensions. The dual differential equation is solved by a method of separation of variable. Based on the integral form of orthotropic elasticity a new orthogonal relationship is proved by using some identical equations. The new orthogonal relationship not only includes the symplectic orthogonal relationship but is also simpler. The physical significance of the new orthogonal relationship is the symmetry representation about an axis z for solutions of the dual equation. The symplectic orthogonal relationship is a generalized relationship but it may be appeared in a strong form with narrow sense in certain condition. This theoretical achievement will provide new effective tools for the research on analytical and finite element solutions to orthotropic elasticity of three-dimensions.展开更多
Aiming at the limitations of psychopathology (PTH), the dual-factor model of mental health (DFM) was proposed as a new mental health concept and methodology under the background of positive psychology trend. In this p...Aiming at the limitations of psychopathology (PTH), the dual-factor model of mental health (DFM) was proposed as a new mental health concept and methodology under the background of positive psychology trend. In this paper we propose giving an overview of DFM, and doubt, criticize, and modify DFM from the perspective of Chinese psychological suzhi research. The available literature from 1983 to 2012 that is related to DFM and concerning psychological suzhi research in the past 20 years has been reviewed. In addition, we also absorbed the idea of positive psychology and Traditional Chinese Medicine (TCM) Constitution theory to develop theoretically the relationship model between psychological suzhi and mental health. The relationship model between psychological suzhi and mental health modifies and transcends PTH and DFM. It will be the new research area of mental health research.展开更多
文摘应用中的各种因素可能造成数据缺失,影响后续任务的分析。因此,数据集缺失值的插补尤为重要。相比原本没有插补的处理,错误的插补值也会对分析造成更严重的偏差。针对这种情况,提出新的采用双重判别器的基于条件生成对抗插补网络(C-GAIN)的缺失值插补算法DDC-GAIN(Dual Discriminator based on C-GAIN)。该算法通过一个辅助判别器辅助主判别器判断预测值的真假,即根据一个样本的全局信息判断这个样本生成的真假,更注重特征之间的关系,以此估算预测值。在4个数据集上与5种经典插补算法进行对比实验,结果表明:同样条件下,DDC-GAIN算法在样本量较大时的均方根误差(RMSE)最低;在Default credit card数据集上缺失率为15%时,DDC-GAIN算法的RMSE比次优算法C-GAIN降低了28.99%。这说明利用辅助判别器帮助主判别器学习特征之间的关系是有效的。
文摘The new orthogonal relationship is generalized for orthotropic elasticity of three-dimensions. The thought of how dual vectors are constructed in a new orthogonal relationship for theory of elasticity is generalized into orthotropic problems. A new dual vector is presented by the dual vector of the symplectic systematic methodology for elasticity that is over again sorted. A dual differential equation is directly obtained by using a mixed variables method. A dual differential matrix to be derived possesses a peculiarity of which principal diagonal sub-matrixes are zero matrixes. As a result of the peculiarity of the dual differential matrix, two independently and symmetrically orthogonal sub-relationships are discovered for orthotropic elasticity of three-dimensions. The dual differential equation is solved by a method of separation of variable. Based on the integral form of orthotropic elasticity a new orthogonal relationship is proved by using some identical equations. The new orthogonal relationship not only includes the symplectic orthogonal relationship but is also simpler. The physical significance of the new orthogonal relationship is the symmetry representation about an axis z for solutions of the dual equation. The symplectic orthogonal relationship is a generalized relationship but it may be appeared in a strong form with narrow sense in certain condition. This theoretical achievement will provide new effective tools for the research on analytical and finite element solutions to orthotropic elasticity of three-dimensions.
文摘Aiming at the limitations of psychopathology (PTH), the dual-factor model of mental health (DFM) was proposed as a new mental health concept and methodology under the background of positive psychology trend. In this paper we propose giving an overview of DFM, and doubt, criticize, and modify DFM from the perspective of Chinese psychological suzhi research. The available literature from 1983 to 2012 that is related to DFM and concerning psychological suzhi research in the past 20 years has been reviewed. In addition, we also absorbed the idea of positive psychology and Traditional Chinese Medicine (TCM) Constitution theory to develop theoretically the relationship model between psychological suzhi and mental health. The relationship model between psychological suzhi and mental health modifies and transcends PTH and DFM. It will be the new research area of mental health research.
