Return to Preprint Details How the Sigmoid Activation Function Causes Vanishing Gradients in Deep Neural Networks Download Download PDF