-Argues that key is transitioning effectively from shallow to deep and residual representations are not necessary(认为关键是有效地从浅到深，而残差表示方法是不必要的)
-Fractal architecture with both shallow and deep paths to output（分形结构是带有浅和深的两种路径到输出）
-Trained with dropping out sub-paths（训练时不带有子路径）
-Full network at test time（测试时带有整个网络）