Deep neural networks (DNN) is a subcategory of machine learning, where the learning ability is modeled after the human brain, and works by structuring learning in layers. In each layer, particular receptors put out a value depending on the data recognition. This goes on until one receptor in the last layer is triggered. This receptor represents the network's decision regarding what the original data represents. By feeding the network with huge amounts of data, the network gets smarter and more accurate.
In special cases and on research/testing basis we use DNN as a part of our solutions.