Going Beyond the 1000-Layer Convolution Network
Author(s): Bartosz Ludwiczuk Originally published on Towards AI. · Introduction· Vanishing gradient issue· Mitigation of the vanishing gradient issue· Training 1000 layer network· Training component analysis· Diving Deeper into Skip Connections· 10000-layer network Mean gradient for 1st layer in all experiments Introduction …