What is DBN (Deep Belief Network)?

Deep belief network (DBN) is a generative graphical model, made out of numerous layers of latent variables, with associations between the layers however not between units inside each layer. DBN are pre-prepared utilizing the Greedy learning algorithm. Greedy learning algorithm utilizes a layer-by-layer method for learning the hierarchical, generative weights. These generative weights decide how factors in one layer rely upon the factors in the layer above.