Probability Density Estimation
Chris Bishop · 1995
Abstract Density estimation can also be applied to unlabelled data (that is data without any class labels) where it has a number of applications. In the context of neural networks it can be applied to the distribution of data in the input space as part of the training process for radial basis function networks (Section 5.9), and to provide a method for validating the outputs of a trained neural network (Bishop, 1994b). In Chapter 6, techniques for density estimation are combined with neural network models to provide a general framework for modelling conditional density functions.