Probability density estimation based on SVM

The problem of probability density estimation can be used in many areas in signal processing, such as regression and classification. In this paper, a density estimation approach based on Support Vector Machine (SVM) was developed. Our algorithm has robust results and sparse solutions compared with Parzen's method. Besides, we used fundamental splines instead of Gaussian kernels in order to further reduce the computation. The simulations show that SVM method for density estimation has a moderately good performance and high convergence speed. Further more, a Bayesian classifier is constructed using the density estimation algorithm.