1 paper · 1 filter
Yucong Liu, Shixing Yu, Tong Lin
In this paper, we develop a novel regularization method for deep neural networks by penalizing the trace of Hessian. This regularizer is motivated by a recent guarantee bound of th…