A neural network is a program that can learn based on data and examples. That is, it does not work according to ready-made rules and algorithms, but writes them itself during training. If you show her a million photographs of cats, she will learn to recognize them in any conditions, poses and costumes.

The trick of the neural network is that the algorithms in it are structured like neurons in the human brain - that is, they are interconnected by synapses and can transmit signals to each other. It is on the strength of these signals that learning depends - for example, in the case of cats, the neural network will form strong connections between neurons that recognize the face and whiskers.

And to make the neuron solve problems even faster, the developers came up with the idea of ​​placing neurons on different layers.