131 citations · 179 across the 26 of their papers we have counts for
1 paper · 2 filters
Kirill Neklyudov, Dmitry Molchanov, Arsenii Ashukha +1
Dropout-based regularization methods can be regarded as injecting random noise with pre-defined magnitude to different parts of the neural network during training. It was recently…