Novel Monte Carlo estimators are proposed to solve both the Tikhonov regularization (TR) and the interpolation problems on graphs. These estimators are based on random spanning forests (RSF), the theoretical properties of which enable to analyze the estimators' theoretical mean and variance. We also show how to perform hyperparameter tuning for these RSF-based estimators. TR is a component in many well-known algorithms, and we show how the proposed estimators can be easily adapted to avoid expensive intermediate steps in generalized semi-supervised learning, label propagation, Newton's method and iteratively reweighted least squares. In the experiments, we illustrate the proposed methods on several problems and provide observations on their run time.
翻译:提议用新蒙特卡洛估计数字来解决Tikhonov的正规化(TR)和图表上的内插问题。这些估计数字基于随机的横贯森林(RSF),其理论特性可以分析估计者的理论平均值和差异。我们还展示了如何对这些基于RSF的估计数字进行超参数调控。TR是许多众所周知的算法中的一个组成部分,我们展示了如何轻而易举地调整拟议的估计数字,以避免在普遍的半监督学习、标签传播、牛顿方法和迭代重标最低平方块中出现昂贵的中间步骤。在实验中,我们介绍了关于若干问题的拟议方法,并提供了运行时间的观察。