Dependence Minimizing Regression with Model Selection for Non-Linear Causal Inference under Non-Gaussian Noise

Dependence Minimizing Regression with Model Selection for Non-Linear Causal Inference under Non-Gaussian Noise
复制标题

DOI：
10.1609/aaai.v24i1.7655
复制
发表时间：
2010-06
期刊：
--
影响因子：
--
通讯作者：
M. Yamada;Masashi Sugiyama
M. Yamada;Masashi Sugiyama
中科院分区：
其他
文献类型：
--
作者：
M. Yamada;Masashi Sugiyama

文献摘要

被引文献

相似文献

加性非高斯噪声模型下的非线性因果关系的发现由于具有很高的灵活性，近年来引起了人们的广泛关注。本文提出了一种新的因果推理算法--最小二乘独立回归(LSIR)。LSIR通过最小化输入和残差之间的平方损失互信息估计器来学习加性噪声模型。与现有方法相比，LSIR的一个显著优点是可以通过交叉验证自然地优化调整参数，如核宽度和正则化参数，从而避免数据依赖的方式过度拟合。通过对真实数据集的实验，我们表明LSIR比最先进的因果推理方法具有更好的性能。
The discovery of non-linear causal relationship under additive non-Gaussian noise models has attracted considerable attention recently because of their high flexibility. In this paper, we propose a novel causal inference algorithm called least-squares independence regression (LSIR). LSIR learns the additive noise model through minimization of an estimator of the squared-loss mutual information between inputs and residuals. A notable advantage of LSIR over existing approaches is that tuning parameters such as the kernel width and the regularization parameter can be naturally optimized by cross-validation, allowing us to avoid overfitting in a data-dependent fashion. Through experiments with real-world datasets, we show that LSIR compares favorably with the state-of-the-art causal inference method.