Training Deep Generative Models in Highly Incomplete Data Scenarios with Prior Regularization | IEEE Conference Publication | IEEE Xplore