A mixture density network can model arbitrary conditional probability distributions in the same manner that conventional neural networks can model arbitrary functions. The 2d case follows the tutorial at http://blog.otoro.net/2015/11/24/mixture-density-networks-with-tensorflow/, and the original paper is "Mixture Density Networks" by C.M. Bishop (1994).
bekaiser / mdn Goto Github PK
View Code? Open in Web Editor NEWMixture Density Networks in Tensorflow