wide_resnet_flc.py

""" Generic Class for Wide ResNet with FLC Pooling
Based on code from https://github.com/yaodongyu/TRADES """
import math
import torch
import torch.nn as nn
import torch.nn.functional as F

from flc_pooling import FLC_Pooling


class BasicBlock(nn.Module):
    def __init__(self, in_planes, out_planes, stride, dropRate=0.0, cutoff=[8,24,4,12,2,6]):
        super(BasicBlock, self).__init__()
        self.bn1 = nn.BatchNorm2d(in_planes)
        self.relu1 = nn.ReLU(inplace=True)
        if stride == 1:
            self.conv1 = nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=1,
                               padding=1, bias=False)
        else:
            self.conv1 = nn.Sequential(
                FLC_Pooling(),
                nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=1,
                               padding=1, bias=False)
            )
        self.bn2 = nn.BatchNorm2d(out_planes)
        self.relu2 = nn.ReLU(inplace=True)
        self.conv2 = nn.Conv2d(out_planes, out_planes, kernel_size=3, stride=1,
                               padding=1, bias=False)
        self.droprate = dropRate
        self.equalInOut = (in_planes == out_planes)

        if not self.equalInOut:
            if stride == 1:
                self.convShortcut = nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=1, padding=0, bias=False)
            else:
                self.convShortcut = nn.Sequential(
                    FLC_Pooling(), 
                    nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=1, padding=0, bias=False)
                    )
        else:
            self.convShortcut = None

    def forward(self, x):
        if not self.equalInOut:
            x = self.relu1(self.bn1(x))
        else:
            out = self.relu1(self.bn1(x))
        out = self.relu2(self.bn2(self.conv1(out if self.equalInOut else x)))
        if self.droprate > 0:
            out = F.dropout(out, p=self.droprate, training=self.training)
        out = self.conv2(out)
        return torch.add(x if self.equalInOut else self.convShortcut(x), out)


class NetworkBlock(nn.Module):
    def __init__(self, nb_layers, in_planes, out_planes, block, stride, dropRate=0.0):
        super(NetworkBlock, self).__init__()
        self.layer = self._make_layer(block, in_planes, out_planes, nb_layers, stride, dropRate)

    def _make_layer(self, block, in_planes, out_planes, nb_layers, stride, dropRate):
        layers = []
        for i in range(int(nb_layers)):
            layers.append(block(i == 0 and in_planes or out_planes, out_planes, i == 0 and stride or 1, dropRate))
        return nn.Sequential(*layers)

    def forward(self, x):
        return self.layer(x)


class WideResNet(nn.Module):
    def __init__(self, depth=28, num_classes=10, widen_factor=10, sub_block1=False, dropRate=0.0, bias_last=True):
        super(WideResNet, self).__init__()
        nChannels = [16, 16 * widen_factor, 32 * widen_factor, 64 * widen_factor]
        assert ((depth - 4) % 6 == 0)
        n = (depth - 4) / 6
        block = BasicBlock
        # 1st conv before any network block
        self.conv1 = nn.Conv2d(3, nChannels[0], kernel_size=3, stride=1,
                               padding=1, bias=False)
        # 1st block
        self.block1 = NetworkBlock(n, nChannels[0], nChannels[1], block, 1, dropRate)
        if sub_block1:
            # 1st sub-block
            self.sub_block1 = NetworkBlock(n, nChannels[0], nChannels[1], block, 1, dropRate)
        # 2nd block
        self.block2 = NetworkBlock(n, nChannels[1], nChannels[2], block, 2, dropRate)
        # 3rd block
        self.block3 = NetworkBlock(n, nChannels[2], nChannels[3], block, 2, dropRate)
        # global average pooling and classifier
        self.bn1 = nn.BatchNorm2d(nChannels[3])
        self.relu = nn.ReLU(inplace=True)
        self.fc = nn.Linear(nChannels[3], num_classes, bias=bias_last)
        self.nChannels = nChannels[3]

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
                m.weight.data.normal_(0, math.sqrt(2. / n))
            elif isinstance(m, nn.BatchNorm2d):
                m.weight.data.fill_(1)
                m.bias.data.zero_()
            elif isinstance(m, nn.Linear) and not m.bias is None:
                m.bias.data.zero_()

    def forward(self, x):
        out = self.conv1(x)
        out = self.block1(out)
        out = self.block2(out)
        out = self.block3(out)
        out = self.relu(self.bn1(out))
        out = F.avg_pool2d(out, 8)
        out = out.view(-1, self.nChannels)
        return self.fc(out)


class WRN_normalized(WideResNet):
    def __init__(self, device='cuda'):
        super(WRN_normalized, self).__init__()
        self.mu = torch.Tensor([0.4914, 0.4822, 0.4465]).float().view(3, 1, 1).to(device)
        self.sigma = torch.Tensor([0.2471, 0.2435, 0.2616]).float().view(3, 1, 1).to(device)
    
    def forward(self, x):
        x = (x - self.mu) / self.sigma
        return super(WRN_normalized, self).forward(x)

def WideResNet2810_normalized(device='cuda'):
    return WRN_normalized(device=device)

def WideResNet2810(device='cuda'):
    return WideResNet()