Activation Functions: Introduce non-linearity into the model, enabling it to learn complex patterns that cannot be captured by linear models. Common activation functions include sigmoid, ReLU (Rectified Linear Unit), and tanh (Hyperbolic Tangent).