Short Definition
A neural network is a machine learning model built from layers of connected computations that transform inputs into outputs.
Intuition
Each layer reshapes the information a little. Stacking layers lets the model learn increasingly useful representations.
Technical Definition
A neural network composes affine transformations, nonlinearities, and other differentiable operations into a trainable function optimized by gradient-based methods.
Example
A neural network can map an image tensor to probabilities for object categories.
Common Misunderstandings
Neural networks are not literal simulations of brains.
More layers do not automatically mean better performance.