Towards Adaptive Residual Network Training: A Neural-ODE Perspective

ICML 2020