Learning threshold neurons via the "edge of stability"

Open in new window