search

Roll920 / ThiNet

caffe model of ICCV'17 paper - ThiNet: A Filter Level Pruning Method for Deep Neural Network Compression https://arxiv.org/abs/1707.06342
144 stars 40 forks source link

Would you please make some comparison with another similar channel pruning work by Megvii? #7

Closed hungsing92 closed 6 years ago

hungsing92 commented 6 years ago

Hi,

Many thanks for your excellent work!!!

I found another ICCV paper similiar with your work.
Would you please make some comment on this paper?

Best, hungsing

Roll920 commented 6 years ago

@hungsing92 This paper has a very similar idea with our ThiNet. The only difference is the method used for channel selection. They proposed a lasso based algorithm, while ours are greedy-based. Hence, these two papers have very similar performances.