Sigmoid
A mathematical activation function that smoothly maps any real-valued number into a strict range between 0 and 1, forming an S-shaped curve.
Think of It Like This
Like a dimmer switch that takes an infinitely powerful electrical current and smoothly caps it to a maximum brightness of 100%.
Historically used to introduce non-linearity in neural networks, it is now primarily restricted to the output layer of binary classification models (to represent probabilities) or as gates in LSTMs. Deep networks avoid using Sigmoid in hidden layers because it suffers heavily from the vanishing gradient problem.