On the Power and Limitations of Random Features for Understanding Neural Networks

Open in new window