What component transforms the weighted sum to introduce non-linearity?

What component transforms the weighted sum to introduce non-linearity?

• A. Input Layer
• B. Output Layer
• C. Activation Function
• D. Weights

Answer: (C)

Non-linearity in neural networks comes from applying an activation function to the neuron’s linear combination of inputs. After you compute the weighted sum plus a bias, the activation function transforms that value in a non-linear way, which lets the network model complex patterns. Without this step, multiple layers would just compose linear transformations, which could be collapsed into one linear map and wouldn’t capture non-linear relationships. Activation functions such as ReLU, sigmoid, or tanh are common examples that introduce this needed non-linearity. The input layer simply passes data forward, and the weights shape the linear combination; the output layer may apply an activation too, but the key element that introduces non-linearity is the activation function.

Non-linearity in neural networks comes from applying an activation function to the neuron’s linear combination of inputs. After you compute the weighted sum plus a bias, the activation function transforms that value in a non-linear way, which lets the network model complex patterns. Without this step, multiple layers would just compose linear transformations, which could be collapsed into one linear map and wouldn’t capture non-linear relationships. Activation functions such as ReLU, sigmoid, or tanh are common examples that introduce this needed non-linearity. The input layer simply passes data forward, and the weights shape the linear combination; the output layer may apply an activation too, but the key element that introduces non-linearity is the activation function.