search

itayhubara / BinaryNet.pytorch

Binarized Neural Network (BNN) for pytorch
504 stars 125 forks source link

About the network inflation factors #7

Open hex0102 opened 6 years ago

hex0102 commented 6 years ago

As I find out in the code, layers in VGG network, ResNet have an inflation factor. Could someone please help clarify this? Why need to inflate the network? Is there a reference to address this question? I also checked the tensorflow repo for the BNN network, there's no inflation factor.

itayhubara commented 6 years ago

you don't have to inflate the network but it improves accuracy. WRPN paper: https://arxiv.org/abs/1709.01134

hex0102 commented 6 years ago

@itayhubara Thank you very much for your quick reply and help. The reference is really useful. I'm checking out this paper.

hex0102 commented 6 years ago

@itayhubara It seems like that for VGG_CIFAR10, the attained accuracy for the inflated network doesn't improve very much (limited to 1%). Are there any reasons to explain this?

itayhubara commented 6 years ago

VGG_CIFAR10 is a very large network to begin with...