一个奇异值神经网络模型及其稳定性分析

A Singular Value Neural Network Model and Its Stability Analysis

提出了一个新的计算实矩阵最大奇异值的神经网络模型,通过求解其微分方程模型的解析解,给出了该模型的渐近稳定性分析,最后给出了数值试验结果,进一步验证了理论分析的正确性.

A novel neural network model for computing the largest singular value of a given real matrix is proposed. By presenting the analytical solution of the differential equations model, asymptotic stability is derived. Finally, a numerical experiment is provided, which further verifies correctness of the theoretical analysis.