All four reviewers note that the paper proposes and investigates an important and very useful class of deep architectures and acknowledge a wide range of potential applications and a number of interesting future extensions. Based on the strongly positive feedback, the paper is recommended to be accepted.