Simple definition
Deep learning trains neural networks with multiple layers to learn patterns and representations from data.
Purpose and use
Deep learning is a machine-learning family using networks with multiple learned processing layers.
How does it work?
Layers can construct successive representations useful for vision, language or other tasks according to architecture and objective.
Example and verification
Before adoption, compare measured improvement with data, compute and maintenance requirements.
Points to watch
A deeper model is not automatically better. Training quality, evaluation and suitability for the problem remain decisive.
In short
Choose a deep network based on demonstrated results and operational constraints.