What is a neural network?

A neural network is a function with adjustable knobs. You give it an input, it produces an output, and training is the process of turning the knobs until the outputs become useful.

Layers of simple steps

Each layer performs a simple calculation and passes its result to the next. Stack enough layers and the network can represent surprisingly rich patterns, from handwriting to language.

Learning from mistakes

During training the network compares its output to the right answer, measures the error and nudges every knob slightly to reduce it. Repeat that millions of times and useful behaviour emerges.

What to remember

Networks do not understand in the human sense. They are statistical tools, which is exactly why evaluation and human oversight matter.