AI can make mistakes. Check important information.
Neural network
Numbers enter on the left, pass through a hidden layer of weighted connections, and leave on the right as a prediction.
Inputs
Each input neuron carries one number describing the example.
Weighted connections
Every input connects to every hidden neuron. Each connection has a weight that scales the signal passing along it.
Hidden layer
Hidden neurons combine their inputs and apply an activation function, letting the network model non-linear patterns.
Output
The output layer combines the hidden activations into the final prediction. Training adjusts the weights to make it accurate.