DBNs are mostly used for unsupervised learning methods like restricted Boltzmann machines (RBMs), Contrastive Divergence (CD) algorithms, and Deep Belief Networks (DBNs). It can be used for classification, dimensionality reduction, and feature extraction.