-
Notifications
You must be signed in to change notification settings - Fork 0
/
pseudo_label_vd.py
197 lines (166 loc) · 7.12 KB
/
pseudo_label_vd.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
import time
import math
from argparse import ArgumentParser
from pathlib import Path
from src.datasets import Cifar10Dataset
from src.datasets import Cutout
from src.resnet import ResNetGenerator
import torch
import torch.nn as nn
import torch.optim as optim
import torchvision.transforms as transforms
import torchvision.models as models
def get_dataloader(args):
transform_train = transforms.Compose([
transforms.RandomCrop(64 if args.data_name == "tiny-imagenet" else 32, padding=4, pad_if_needed=True),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
Cutout(1, 3),
])
transform_test = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
])
trainset = Cifar10Dataset(args.clean, transform_train, train=True, show_backdoor=False)
testset_clean = Cifar10Dataset(args.clean, transform_test, train=False, show_backdoor=False)
trainset_pseudo = Cifar10Dataset(args.data, transform_test, train=True, show_backdoor=True)
train_loader = torch.utils.data.DataLoader(trainset, batch_size=args.batch_size, shuffle=True, num_workers=8)
test_clean_loader = torch.utils.data.DataLoader(testset_clean, batch_size=100, shuffle=False, num_workers=8)
train_pseudo_loader = torch.utils.data.DataLoader(trainset_pseudo, batch_size=100, shuffle=False, num_workers=8)
return train_loader, test_clean_loader, train_pseudo_loader
def train(model, train_loader, criterion, optimizer, device):
model.train()
acc_cnt = 0
all_cnt = 0
loss_log = 0
for i, (image, label) in enumerate(train_loader):
image = image.to(device)
label = label.to(device)
logits = model(image)
loss = criterion(logits, label)
optimizer.zero_grad()
loss.backward()
optimizer.step()
acc_cnt += (logits.detach().max(1)[1] == label).sum()
all_cnt += len(label)
loss_log += loss.detach() * len(label)
train_acc = acc_cnt / all_cnt * 100
loss = loss_log / acc_cnt
return train_acc, loss
def test(model, test_clean_loader, device):
model.eval()
with torch.no_grad():
acc_cnt = 0
acc_total = 0
for i, (image, label) in enumerate(test_clean_loader):
image = image.to(device)
label = label.to(device)
logits = model(image)
_, pred = logits.max(1)
acc_cnt += (pred == label).int().sum()
acc_total += len(label)
dev_acc = acc_cnt / acc_total * 100
return dev_acc
def update_pseudo_label(model, train_pseudo_loader, device):
model.eval()
print("Updating pseudo labels...")
pseudo_labels = []
true_labels = []
backdoor = []
with torch.no_grad():
for i, (image, _, true_label, bd, _) in enumerate(train_pseudo_loader):
image = image.to(device)
true_label = true_label.to(device)
bd = bd.to(device)
image[:, :, :2, :2] = 0
logits = model(image)
_, pred = logits.max(1)
pseudo_labels.append(pred)
true_labels.append(true_label)
backdoor.append(bd)
pseudo_labels = torch.cat(pseudo_labels)
true_labels = torch.cat(true_labels)
backdoor = torch.cat(backdoor)
print("Pseudo Label Acc: {:.2f}%".format(100 * (pseudo_labels == true_labels).sum() / len(backdoor)))
return pseudo_labels
def main(args):
print("Running")
# get data
train_loader, test_clean_loader, train_pseudo_loader = get_dataloader(args)
# get model
model = ResNetGenerator(args.arch, num_splits=1, num_classes=args.num_classes)
model = model.to(args.device)
if args.model != "":
state_dict = torch.load(args.model)
model.load_state_dict(state_dict)
else:
if args.freeze > 0:
freeze(model)
# get optimizer and criterion
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=args.lr, momentum=0.9, weight_decay=args.weight_decay)
epoch = 0
while epoch < args.epochs:
tik = time.time()
if epoch == args.freeze:
unfreeze(model)
adjust_lr(optimizer, epoch, args)
train_acc, train_loss = train(model, train_loader, criterion, optimizer, args.device)
dev_acc = test(model, test_clean_loader, args.device)
tok = time.time()
print("Epoch: {} | Acc: {:.2f}% | Loss: {:.3f} | Dev Acc: {:.2f}% | Time: {:.2f}".format(
epoch, train_acc, train_loss, dev_acc, tok - tik))
if epoch + 1 == args.epochs:
torch.save(model.state_dict(), args.model_save_dir / "model_lite.pt")
epoch += 1
pseudo_label = update_pseudo_label(model, train_pseudo_loader, args.device)
# torch.save(pseudo_label.cpu(), f"pseudo_label/{args.data_name}_{args.attack}_vd")
def freeze(model):
print("==> Freeze feature extractor")
for name, param in model.named_parameters():
if name not in ['fc.weight', 'fc.bias', 'module.fc.weight', 'module.fc.bias']:
param.requires_grad = False
def unfreeze(model):
print("==> Unfreeze feature extractor")
for name, param in model.named_parameters():
if name not in ['fc.weight', 'fc.bias', 'module.fc.weight', 'module.fc.bias']:
param.requires_grad = True
def adjust_lr(optimizer, epoch, args):
if args.data_name == "cifar10":
if epoch < 20: # cifar10
lr = 0.01
elif epoch < 60:
lr = 0.001
else:
lr = 0.0001
else:
if epoch < args.freeze: # tiny-imagenet
lr = args.lr * 10
else:
lr = 0.5 * (1 + math.cos(math.pi * (epoch - args.freeze + 1) / (args.epochs - args.freeze + 1))) * args.lr
for param_group in optimizer.param_groups:
param_group['lr'] = lr
if __name__ == "__main__":
parser = ArgumentParser()
parser.add_argument("--data", type=str, default="cifar10")
parser.add_argument("--arch", type=str, default="resnet-18")
parser.add_argument("--attack", type=str, default="badnets10")
parser.add_argument("--model", type=str, default="")
parser.add_argument("--batch-size", type=int, default=64)
parser.add_argument("--lr", type=float, default=0.1)
parser.add_argument("--weight-decay", type=float, default=1e-4)
parser.add_argument("--epochs", type=int, default=100)
parser.add_argument("--freeze", type=int, default=0)
args = parser.parse_args()
args.device = "cuda" if torch.cuda.is_available() else "cpu"
args.num_classes = 200 if args.data == "tiny-imagenet" else 10
args.data_name = args.data
args.data = Path("datasets") / args.data_name / args.attack
args.clean = Path("datasets") / args.data_name / "clean_lite"
args.model_save_dir = Path("checkpoints") / f"{args.data_name}_clean_lite"
args.model_save_dir.mkdir(exist_ok=True)
args.label_save_dir = Path("pseudo_label")
args.label_save_dir.mkdir(exist_ok=True)
print(args)
main(args)