智能反射面(intelligent reflecting surface,IRS)辅助通信系统的信道维度较高,现有的信道估计方法须使用大量导频才能得到准确的信道矩阵.针对这一问题,提出了一种基于混合损失的残差生成对抗网络(hybrid loss based residual generati...智能反射面(intelligent reflecting surface,IRS)辅助通信系统的信道维度较高,现有的信道估计方法须使用大量导频才能得到准确的信道矩阵.针对这一问题,提出了一种基于混合损失的残差生成对抗网络(hybrid loss based residual generative adversarial network,H-ResGAN)模型.H-ResGAN使用多个残差块来加深网络,可以充分提取信道特征,减缓梯度消失问题.同时,采用条件最小二乘损失和L1损失相结合的混合损失作为目标函数来提高训练的稳定性.仿真实验证明:H-ResGAN对环境噪声更具鲁棒性,估计误差显著低于传统方法;与传统的估计算法相比,H-ResGAN仅须发送少量导频就能获得准确的估计结果.展开更多
Conventional artificial neural networks used to solve electrical resistivity imaging （ERI） inversion problem suffer from overfitting and local minima. To solve these problems, we propose to use a pruning Bayesian ne...Conventional artificial neural networks used to solve electrical resistivity imaging （ERI） inversion problem suffer from overfitting and local minima. To solve these problems, we propose to use a pruning Bayesian neural network （PBNN） nonlinear inversion method and a sample design method based on the K-medoids clustering algorithm. In the sample design method, the training samples of the neural network are designed according to the prior information provided by the K-medoids clustering results; thus, the training process of the neural network is well guided. The proposed PBNN, based on Bayesian regularization, is used to select the hidden layer structure by assessing the effect of each hidden neuron to the inversion results. Then, the hyperparameter αk, which is based on the generalized mean, is chosen to guide the pruning process according to the prior distribution of the training samples under the small-sample condition. The proposed algorithm is more efficient than other common adaptive regularization methods in geophysics. The inversion of synthetic data and field data suggests that the proposed method suppresses the noise in the neural network training stage and enhances the generalization. The inversion results with the proposed method are better than those of the BPNN, RBFNN, and RRBFNN inversion methods as well as the conventional least squares inversion.展开更多
To improve the global search ability and imaging quality of electrical resistivity imaging(ERI) inversion, a two-stage learning ICPSO algorithm of radial basis function neural network(RBFNN) based on information crite...To improve the global search ability and imaging quality of electrical resistivity imaging(ERI) inversion, a two-stage learning ICPSO algorithm of radial basis function neural network(RBFNN) based on information criterion(IC) and particle swarm optimization(PSO) is presented. In the proposed method, IC is applied to obtain the hidden layer structure by calculating the optimal IC value automatically and PSO algorithm is used to optimize the centers and widths of the radial basis functions in the hidden layer. Meanwhile, impacts of different information criteria to the inversion results are compared, and an implementation of the proposed ICPSO algorithm is given. The optimized neural network has one hidden layer with 261 nodes selected by AKAIKE's information criterion(AIC) and it is trained on 32 data sets and tested on another 8 synthetic data sets. Two complex synthetic examples are used to verify the feasibility and effectiveness of the proposed method with two learning stages. The results show that the proposed method has better performance and higher imaging quality than three-layer and four-layer back propagation neural networks(BPNNs) and traditional least square(LS) inversion.展开更多
文摘智能反射面(intelligent reflecting surface,IRS)辅助通信系统的信道维度较高,现有的信道估计方法须使用大量导频才能得到准确的信道矩阵.针对这一问题,提出了一种基于混合损失的残差生成对抗网络(hybrid loss based residual generative adversarial network,H-ResGAN)模型.H-ResGAN使用多个残差块来加深网络,可以充分提取信道特征,减缓梯度消失问题.同时,采用条件最小二乘损失和L1损失相结合的混合损失作为目标函数来提高训练的稳定性.仿真实验证明:H-ResGAN对环境噪声更具鲁棒性,估计误差显著低于传统方法;与传统的估计算法相比,H-ResGAN仅须发送少量导频就能获得准确的估计结果.
基金supported by the National Natural Science Foundation of China(Grant No.41374118)the Research Fund for the Higher Education Doctoral Program of China(Grant No.20120162110015)+3 种基金the China Postdoctoral Science Foundation(Grant No.2015M580700)the Hunan Provincial Natural Science Foundation,the China(Grant No.2016JJ3086)the Hunan Provincial Science and Technology Program,China(Grant No.2015JC3067)the Scientific Research Fund of Hunan Provincial Education Department,China(Grant No.15B138)
文摘Conventional artificial neural networks used to solve electrical resistivity imaging （ERI） inversion problem suffer from overfitting and local minima. To solve these problems, we propose to use a pruning Bayesian neural network （PBNN） nonlinear inversion method and a sample design method based on the K-medoids clustering algorithm. In the sample design method, the training samples of the neural network are designed according to the prior information provided by the K-medoids clustering results; thus, the training process of the neural network is well guided. The proposed PBNN, based on Bayesian regularization, is used to select the hidden layer structure by assessing the effect of each hidden neuron to the inversion results. Then, the hyperparameter αk, which is based on the generalized mean, is chosen to guide the pruning process according to the prior distribution of the training samples under the small-sample condition. The proposed algorithm is more efficient than other common adaptive regularization methods in geophysics. The inversion of synthetic data and field data suggests that the proposed method suppresses the noise in the neural network training stage and enhances the generalization. The inversion results with the proposed method are better than those of the BPNN, RBFNN, and RRBFNN inversion methods as well as the conventional least squares inversion.
基金Project(41374118)supported by the National Natural Science Foundation,ChinaProject(20120162110015)supported by Research Fund for the Doctoral Program of Higher Education,China+3 种基金Project(2015M580700)supported by the China Postdoctoral Science Foundation,ChinaProject(2016JJ3086)supported by the Hunan Provincial Natural Science Foundation,ChinaProject(2015JC3067)supported by the Hunan Provincial Science and Technology Program,ChinaProject(15B138)supported by the Scientific Research Fund of Hunan Provincial Education Department,China
文摘To improve the global search ability and imaging quality of electrical resistivity imaging(ERI) inversion, a two-stage learning ICPSO algorithm of radial basis function neural network(RBFNN) based on information criterion(IC) and particle swarm optimization(PSO) is presented. In the proposed method, IC is applied to obtain the hidden layer structure by calculating the optimal IC value automatically and PSO algorithm is used to optimize the centers and widths of the radial basis functions in the hidden layer. Meanwhile, impacts of different information criteria to the inversion results are compared, and an implementation of the proposed ICPSO algorithm is given. The optimized neural network has one hidden layer with 261 nodes selected by AKAIKE's information criterion(AIC) and it is trained on 32 data sets and tested on another 8 synthetic data sets. Two complex synthetic examples are used to verify the feasibility and effectiveness of the proposed method with two learning stages. The results show that the proposed method has better performance and higher imaging quality than three-layer and four-layer back propagation neural networks(BPNNs) and traditional least square(LS) inversion.
