Mixture of nonlinear models: a Bayesian fit for Principal Curves
Pedro Delicado, Marcelo Smrekar · IEEE International Conference on Neural Networks/IEEE ... International Conference on Neural Networks · 2007
Principal curves are smooth parametric curves passing through the "middle" of a non-elliptical multivariate data set. We model the probability distribution of this kind of data as a mixture of simple nonlinear models and use MCMC techniques to fit the mixture model.