Learning threshold neurons via the "edge of stability "

Neural Information Processing Systems 

Large step sizes are necessary to learn the "threshold neuron" of a ReLU network

Similar Docs  Excel Report  more

TitleSimilaritySource
None found