Instead of organizing data to run through predefined equations, deep neural networks train the algorithm to learn on its own by recognizing patterns using many layers of processing. Deep neural networks, also referred to as deeplearning networks, are neural networks with additional hidden layers.
Each layer of nodes trains on a set of features based on the previous layer’s output. Hence, progression through each layer results in increasingly complex layers of features that are aggregated, recombined information learned from the previous layer. This characteristic makes deep-learning networks able to identify highly complex nonlinear patterns with large volumes of data and dimensions (Press 2017).
On the flip side, deep neural networks are also prone to overfitting. Overfitting may be reduced by validating a new model on an out-of-sample data set. Other popular techniques include support vector machines