1 paper
Amin Ghiasi, Ali Shafahi, Reza Ardekani
We propose adaptive weight decay, which automatically tunes the hyper-parameter for weight decay during each training iteration. For classification problems, we propose changing th…