Maximum-and-Concatenation Networks

Xingyu Xie, Hao Kong, Jianlong Wu +3

While successful in many fields, deep neural networks (DNNs) still suffer from some open problems such as bad local minima and unsatisfactory generalization performance. In this wo…