Constraint-based learning for non-parametric continuous bayesian networks

Modeling high-dimensional multivariate distributions is a computationally challenging task. In the discrete case, Bayesian networks have been successfully used to reduce complexity and simplify problem. However, they lack of general model for continuous variables. order overcome this problem, Elidan (2010) proposed copula that parametrizes using functions. We propose new learning algorithm based on PC conditional independence test by Bouezmarni et al. (2009). This being non-parametric, no assumptions are made allowing it be as possible. compared generated data with parametric method proves better results.