1 paper
Zekun Shi, Zheyuan Hu, Min Lin +1
Optimizing neural networks with loss that contain high-dimensional and high-order differential operators is expensive to evaluate with back-propagation due to O(dk)…