The sigmoid function is defined by the following equation:

It is used in neural networks because it’s differentiable, and this is necessary for back propagation, and its co-domain is , which can be used as probabilities.

mlai