1 paper · 1 filter
Ruoran Xu, Borong She, Xiaobo Jin +1
Deep learning optimization relies heavily on the assumption of smooth loss landscapes, a condition systematically violated by modern architectures due to non-smooth components such…