Which component applies a non-linear transformation to a neuron's weighted input?

Which component applies a non-linear transformation to a neuron's weighted input?

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

Answer: (B)

Non-linearity is introduced by the activation function. After the neuron sums its inputs each weighted by their connections and adds the bias, the activation function is applied to that weighted input. This step is what bends the relationship and allows the network to model complex patterns. Without this non-linear transformation, multiple layers would just compose linear operations into a single linear function, no matter how many layers there are. The weights and bias simply form the linear part of the calculation, adjusting scale and offset, and the term “output layer” is not the element that injects non-linearity by itself. The activation function is the component that makes the output respond in a non-linear way.

Non-linearity is introduced by the activation function. After the neuron sums its inputs each weighted by their connections and adds the bias, the activation function is applied to that weighted input. This step is what bends the relationship and allows the network to model complex patterns. Without this non-linear transformation, multiple layers would just compose linear operations into a single linear function, no matter how many layers there are. The weights and bias simply form the linear part of the calculation, adjusting scale and offset, and the term “output layer” is not the element that injects non-linearity by itself. The activation function is the component that makes the output respond in a non-linear way.