Rethinking Layer-wise Feature Amounts in Convolutional Neural Network Architectures

Open in new window