Residual slayer possible?

[zhaxidele]

Hi, guys,

Slayer works impressively on datasets that have shallow features, but anybody knows if there is somebody who tried going deeper with slayer using residual block? I tried two CNN and two FC on my dataset, it gets only around 45% accuracy for classification, when I give one more CNN, it gets worse with almost no decrease in loss. With ResNet of 21 layers CNN, my data set got over 90% accuracy. So I want to using residual form of SLAYER, and anybody has any resource of this? papers, repositories, etc. ?

Thanks : )