1 code implementation • 6 Jun 2022 • Rhea Chowers, Yair Weiss
It has previously been reported that the representation that is learned in the first layer of deep Convolutional Neural Networks (CNNs) is highly consistent across initializations and architectures.
