Neural network
Also called neural net
A kind of machine-learning model made of layers of simple connected units, whose connection strengths are tuned during training.
A neural network is a stack of layers. Each layer is a large number of small units; each unit takes in numbers, weighs them, adds them up and passes a result to the next layer. The weights are what the network learns. Training nudges billions of them, a tiny amount at a time, until the outputs come out right on the examples.
Nobody decides what any single unit does. Useful behaviour emerges from all of them together, which is why these systems are so capable and why it's hard to say exactly why one gave a particular answer. The 2024 Nobel Prize in Physics went to 2 of the researchers whose work made them possible.
For example
The model that turns your spoken "remind me to call Mum at six" into text is a neural network: sound goes in as numbers, passes through layers, and words come out.
Put it to use
Where you'll meet neural network in ChatGPT, Gemini and Claude, how to use it there, and the mix-up most people make.
Daily Bytes Plus is in the Android app.