提出基于对称正定矩阵流形（SPD）的图神经网络模型SPDGNN，利用Log-Cholesky度量直接在SPD流形上操作，避免频繁映射提升效率，开发专用邻居聚合和非线性模块，构建多类逻辑回归框架，实验证明在层次结构数据分类任务中优于传统欧几里得和超曲面GNN。 非欧几里 ...

We propose new regression models for parameterizing covariance structures in longitudinal data analysis. Using a novel Cholesky factor, the entries in this decomposition have a moving average and ...

Kernel ridge regression (KRR) is a regression technique for predicting a single numeric value and can deliver high accuracy for complex, non-linear data. KRR combines a kernel function (most commonly ...