/
adv_loss.py
256 lines (199 loc) · 7.34 KB
/
adv_loss.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
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.autograd import Variable
def squared_l2_norm(x):
flattened = x.view(x.unsqueeze(0).shape[0], -1)
return (flattened ** 2).sum(1)
def l2_norm(x):
return squared_l2_norm(x).sqrt()
def adv_loss_pgd(model,
x_natural,
y,
optimizer,
step_size=0.003,
epsilon=0.031,
perturb_steps=10,
beta=1.0,
distance='l_inf'):
model.eval()
batch_size = len(x_natural)
# generate adversarial example
x_adv = x_natural.detach() + 0.001 * torch.randn(x_natural.shape).cuda().detach()
if distance == 'l_inf':
for _ in range(perturb_steps):
x_adv.requires_grad_()
with torch.enable_grad():
loss_adv = F.cross_entropy(model(x_adv), y)
grad = torch.autograd.grad(loss_adv, [x_adv])[0]
x_adv = x_adv.detach() + step_size * torch.sign(grad.detach())
x_adv = torch.min(torch.max(x_adv, x_natural - epsilon), x_natural + epsilon)
x_adv = torch.clamp(x_adv, 0.0, 1.0)
elif distance == 'l_2':
for _ in range(perturb_steps):
x_adv.requires_grad_()
with torch.enable_grad():
loss_adv = F.cross_entropy(model(x_adv), y)
grad = torch.autograd.grad(loss_adv, [x_adv])[0]
for idx_batch in range(batch_size):
grad_idx = grad[idx_batch]
grad_idx_norm = l2_norm(grad_idx)
grad_idx /= (grad_idx_norm + 1e-8)
x_adv[idx_batch] = x_adv[idx_batch].detach() + step_size * grad_idx
eta_x_adv = x_adv[idx_batch] - x_natural[idx_batch]
norm_eta = l2_norm(eta_x_adv)
if norm_eta > epsilon:
eta_x_adv = eta_x_adv * epsilon / l2_norm(eta_x_adv)
x_adv[idx_batch] = x_natural[idx_batch] + eta_x_adv
x_adv = torch.clamp(x_adv, 0.0, 1.0)
else:
x_adv = torch.clamp(x_adv, 0.0, 1.0)
model.train()
x_adv = Variable(torch.clamp(x_adv, 0.0, 1.0), requires_grad=False)
# zero gradient
optimizer.zero_grad()
# calculate robust loss
logits = model(x_natural)
loss_natural = F.cross_entropy(logits, y)
loss_robust = F.cross_entropy(model(x_adv), y)
'''loss_robust = (1.0 / batch_size) * criterion_kl(F.log_softmax(model(x_adv), dim=1),
F.softmax(model(x_natural), dim=1))'''
loss = (loss_natural + beta * loss_robust)/(1.0 + beta)
return loss
#==================== L2L_0 Loss ============================
def adv_loss_l2l_0(model,
attacker,
x_natural,
y,
optimizer,
optimizer_att,
beta=1.0,
for_attacker = 0):
if for_attacker == 0:
model.train()
attacker.eval()
else:
model.eval()
attacker.train()
batch_size = len(x_natural)
advinput = x_natural
# generate adversarial example
perturbation = attacker(advinput)
x_adv = x_natural + perturbation
x_adv = torch.clamp(x_adv, 0.0, 1.0)
optimizer.zero_grad()
optimizer_att.zero_grad()
loss_robust = F.cross_entropy(model(x_adv), y)
loss = loss_robust
return loss
#=================== L2L_1 Loss ===========================
def adv_loss_l2l_1(model,
attacker,
x_natural,
y,
optimizer,
optimizer_att,
beta=1.0,
epsilon = 0.031,
for_attacker = 0):
if for_attacker == 0:
model.train()
attacker.eval()
else:
model.eval()
attacker.train()
batch_size = len(x_natural)
x_natural.requires_grad_()
with torch.enable_grad():
loss_natural = F.cross_entropy(model(x_natural), y)
grad = torch.autograd.grad(loss_natural, [x_natural])[0]
advinput = torch.cat([x_natural,1.0*(grad/grad.abs().max())], 1).detach()
# generate adversarial example
perturbation = attacker(advinput)
x_adv = x_natural + perturbation
x_adv = torch.clamp(x_adv, 0.0, 1.0)
optimizer.zero_grad()
optimizer_att.zero_grad()
loss_robust = F.cross_entropy(model(x_adv), y)
loss = loss_robust
return loss
#=================== L2L_2 Loss =============================
def adv_loss_l2l_2(model,
attacker,
x_natural,
y,
optimizer,
optimizer_att,
epsilon=0.031,
beta=1.0,
for_attacker = 0):
if for_attacker == 0:
model.train()
attacker.eval()
else:
model.eval()
attacker.train()
batch_size = len(x_natural)
x_natural.requires_grad_()
with torch.enable_grad():
loss_natural = F.cross_entropy(model(x_natural), y)
grad = torch.autograd.grad(loss_natural, [x_natural])[0]
advinput = torch.cat([x_natural,1.0*(grad/grad.abs().max())], 1).detach()
# generate adversarial example
perturbation = attacker(advinput)
x_adv = x_natural + perturbation
x_adv = torch.clamp(x_adv, 0.0, 1.0)
x_adv.requires_grad_()
with torch.enable_grad():
loss_adv = F.cross_entropy(model(x_adv), y)
grad_adv = torch.autograd.grad(loss_adv, [x_adv])[0]
advinput_1 = torch.cat([x_adv,1.0*(grad_adv/grad_adv.abs().max())], 1)
perturbation_1 = attacker(advinput_1)
perturbation_total = perturbation + perturbation_1
perturbation_total = torch.clamp(perturbation_total, -epsilon, epsilon)
x_adv_final = x_natural + perturbation_total
x_adv_final = torch.clamp(x_adv_final, 0.0, 1.0)
optimizer.zero_grad()
optimizer_att.zero_grad()
loss_robust = F.cross_entropy(model(x_adv_final), y)
loss = loss_robust
return loss
#=================== L2L_k Loss =============================
def adv_loss_l2l_k(k,
model,
attacker,
x_natural,
y,
optimizer,
optimizer_att,
step_size = 0.007,
epsilon=0.031,
beta=1.0,
for_attacker = 0):
if for_attacker == 0:
model.train()
attacker.eval()
else:
model.eval()
attacker.train()
batch_size = len(x_natural)
x_adv = Variable(x_natural.data, requires_grad = True)
for _ in range(k):
with torch.enable_grad():
loss_adv = F.cross_entropy(model(x_adv), y)
grad = torch.autograd.grad(loss_adv, [x_adv])[0]
advinput = torch.cat([x_adv,1.0*(grad/grad.abs().max())], 1).detach()
# generate adversarial example
perturbation = attacker(advinput)
x_adv = x_adv + perturbation*step_size/epsilon
x_adv = torch.clamp(x_adv, 0.0, 1.0)
perturbation_total = x_adv - x_natural
perturbation_total = torch.clamp(perturbation_total, -epsilon, epsilon)
x_adv_final = x_natural + perturbation_total
x_adv_final = torch.clamp(x_adv_final, 0.0, 1.0)
optimizer.zero_grad()
optimizer_att.zero_grad()
# calculate robust loss
loss_robust = F.cross_entropy(model(x_adv_final), y)
loss = loss_robust
return loss