Towards Understanding Generalization in DP-GD: A Case Study in Training Two-Layer CNNs

Open in new window