Least-Squares Independence Regression for Non-Linear Causal Inference Under Non-Gaussian Noise
Abstract
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 the minimization of an estimator of the squared-loss mutual information between inputs and residuals. A notable advantage of LSIR 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 a state-of-the-art causal inference method.
Cite
Text
Yamada et al. "Least-Squares Independence Regression for Non-Linear Causal Inference Under Non-Gaussian Noise." Machine Learning, 2014. doi:10.1007/S10994-013-5423-YMarkdown
[Yamada et al. "Least-Squares Independence Regression for Non-Linear Causal Inference Under Non-Gaussian Noise." Machine Learning, 2014.](https://mlanthology.org/mlj/2014/yamada2014mlj-leastsquares/) doi:10.1007/S10994-013-5423-YBibTeX
@article{yamada2014mlj-leastsquares,
title = {{Least-Squares Independence Regression for Non-Linear Causal Inference Under Non-Gaussian Noise}},
author = {Yamada, Makoto and Sugiyama, Masashi and Sese, Jun},
journal = {Machine Learning},
year = {2014},
pages = {249-267},
doi = {10.1007/S10994-013-5423-Y},
volume = {96},
url = {https://mlanthology.org/mlj/2014/yamada2014mlj-leastsquares/}
}