Deep Belief Networks

What are Deep Belief Networks?

Deep belief networks (DBNs) are generative deep models built by stacking restricted Boltzmann machines, historically important for pre-training deep architectures before end-to-end backpropagation became dominant.

They helped establish deep learning practice; modern work more often uses deep neural nets, transformers, and other architectures for most tasks.

Authoritative reference: Wikipedia: Deep belief network