In Depth
Neural networks consist of an input layer, one or more hidden layers, and an output layer. Deep neural networks — those with many hidden layers — are capable of learning hierarchical representations, such as edges → shapes → objects in images. They are the core building block of modern machine learning.