How do neural networks introduce nonlinearity in their functioning?

Group of answer choices

The activation function in neural networks is always linear, maintaining a linear relationship between inputs and outputs.

A nonlinear activation function, such as the sigmoid function, processes the scalar value obtained from the linear combination, introducing nonlinearity.

Neural networks use a linear combination of node inputs and connection weights, producing a linear output.

The combination function ensures a linear relationship between input and output in neural networks.