-
Notifications
You must be signed in to change notification settings - Fork 47
/
test_all.py
264 lines (247 loc) · 12.7 KB
/
test_all.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
256
257
258
259
260
261
262
263
## test_attack.py -- sample code to test attack procedure
##
## Copyright (C) IBM Corp, 2017-2018
## Copyright (C) 2017, Huan Zhang <[email protected]>.
## Copyright (C) 2016, Nicholas Carlini <[email protected]>.
##
## This program is licenced under the BSD 2-Clause licence,
## contained in the LICENCE file in this directory.
import os
import sys
import tensorflow as tf
import numpy as np
import random
import time
from setup_cifar import CIFAR, CIFARModel
from setup_mnist import MNIST, MNISTModel
from setup_inception import ImageNet, InceptionModel
from l2_attack import CarliniL2
from l2_attack_black import BlackBoxL2
from PIL import Image
def show(img, name = "output.png"):
"""
Show MNSIT digits in the console.
"""
np.save(name, img)
fig = np.around((img + 0.5)*255)
fig = fig.astype(np.uint8).squeeze()
pic = Image.fromarray(fig)
# pic.resize((512,512), resample=PIL.Image.BICUBIC)
pic.save(name)
remap = " .*#"+"#"*100
img = (img.flatten()+.5)*3
if len(img) != 784: return
print("START")
for i in range(28):
print("".join([remap[int(round(x))] for x in img[i*28:i*28+28]]))
def generate_data(data, samples, targeted=True, start=0, inception=False):
"""
Generate the input data to the attack algorithm.
data: the images to attack
samples: number of samples to use
targeted: if true, construct targeted attacks, otherwise untargeted attacks
start: offset into data to use
inception: if targeted and inception, randomly sample 100 targets intead of 1000
"""
inputs = []
targets = []
labels = []
true_ids = []
for i in range(samples):
if targeted:
if inception:
# for inception, randomly choose 10 target classes
seq = np.random.choice(range(1,1001), 10)
# seq = [580] # grand piano
else:
# for CIFAR and MNIST, generate all target classes
seq = range(data.test_labels.shape[1])
# print ('image label:', np.argmax(data.test_labels[start+i]))
for j in seq:
# skip the original image label
if (j == np.argmax(data.test_labels[start+i])) and (inception == False):
continue
inputs.append(data.test_data[start+i])
targets.append(np.eye(data.test_labels.shape[1])[j])
labels.append(data.test_labels[start+i])
true_ids.append(start+i)
else:
inputs.append(data.test_data[start+i])
targets.append(data.test_labels[start+i])
labels.append(data.test_labels[start+i])
true_ids.append(start+i)
inputs = np.array(inputs)
targets = np.array(targets)
labels = np.array(labels)
true_ids = np.array(true_ids)
return inputs, targets, labels, true_ids
def main(args):
with tf.Session() as sess:
use_log = not args['use_zvalue']
is_inception = args['dataset'] == "imagenet"
# load network
print('Loading model', args['dataset'])
if args['dataset'] == "mnist":
data, model = MNIST(), MNISTModel("models/mnist", sess, use_log)
# data, model = MNIST(), MNISTModel("models/mnist-distilled-100", sess, use_log)
elif args['dataset'] == "cifar10":
data, model = CIFAR(), CIFARModel("models/cifar", sess, use_log)
# data, model = CIFAR(), CIFARModel("models/cifar-distilled-100", sess, use_log)
elif args['dataset'] == "imagenet":
data, model = ImageNet(), InceptionModel(sess, use_log)
print('Done...')
if args['numimg'] == 0:
args['numimg'] = len(data.test_labels) - args['firstimg']
print('Using', args['numimg'], 'test images')
# load attack module
if args['attack'] == "white":
# batch size 1, optimize on 1 image at a time, rather than optimizing images jointly
attack = CarliniL2(sess, model, batch_size=1, max_iterations=args['maxiter'], print_every=args['print_every'],
early_stop_iters=args['early_stop_iters'], confidence=0, learning_rate = args['lr'], initial_const=args['init_const'],
binary_search_steps=args['binary_steps'], targeted=not args['untargeted'], use_log=use_log,
adam_beta1=args['adam_beta1'], adam_beta2=args['adam_beta2'])
else:
# batch size 128, optimize on 128 coordinates of a single image
attack = BlackBoxL2(sess, model, batch_size=128, max_iterations=args['maxiter'], print_every=args['print_every'],
early_stop_iters=args['early_stop_iters'], confidence=0, learning_rate = args['lr'], initial_const=args['init_const'],
binary_search_steps=args['binary_steps'], targeted=not args['untargeted'], use_log=use_log, use_tanh=args['use_tanh'],
use_resize=args['use_resize'], adam_beta1=args['adam_beta1'], adam_beta2=args['adam_beta2'], reset_adam_after_found=args['reset_adam'],
solver=args['solver'], save_ckpts=args['save_ckpts'], load_checkpoint=args['load_ckpt'], start_iter=args['start_iter'],
init_size=args['init_size'], use_importance=not args['uniform'])
random.seed(args['seed'])
np.random.seed(args['seed'])
print('Generate data')
all_inputs, all_targets, all_labels, all_true_ids = generate_data(data, samples=args['numimg'], targeted=not args['untargeted'],
start=args['firstimg'], inception=is_inception)
print('Done...')
os.system("mkdir -p {}/{}".format(args['save'], args['dataset']))
img_no = 0
total_success = 0
l2_total = 0.0
for i in range(all_true_ids.size):
inputs = all_inputs[i:i+1]
targets = all_targets[i:i+1]
labels = all_labels[i:i+1]
print("true labels:", np.argmax(labels), labels)
print("target:", np.argmax(targets), targets)
# test if the image is correctly classified
original_predict = model.model.predict(inputs)
original_predict = np.squeeze(original_predict)
original_prob = np.sort(original_predict)
original_class = np.argsort(original_predict)
print("original probabilities:", original_prob[-1:-6:-1])
print("original classification:", original_class[-1:-6:-1])
print("original probabilities (most unlikely):", original_prob[:6])
print("original classification (most unlikely):", original_class[:6])
if original_class[-1] != np.argmax(labels):
print("skip wrongly classified image no. {}, original class {}, classified as {}".format(i, np.argmax(labels), original_class[-1]))
continue
img_no += 1
timestart = time.time()
adv, const = attack.attack_batch(inputs, targets)
if type(const) is list:
const = const[0]
if len(adv.shape) == 3:
adv = adv.reshape((1,) + adv.shape)
timeend = time.time()
l2_distortion = np.sum((adv-inputs)**2)**.5
adversarial_predict = model.model.predict(adv)
adversarial_predict = np.squeeze(adversarial_predict)
adversarial_prob = np.sort(adversarial_predict)
adversarial_class = np.argsort(adversarial_predict)
print("adversarial probabilities:", adversarial_prob[-1:-6:-1])
print("adversarial classification:", adversarial_class[-1:-6:-1])
success = False
if args['untargeted']:
if adversarial_class[-1] != original_class[-1]:
success = True
else:
if adversarial_class[-1] == np.argmax(targets):
success = True
if l2_distortion > 20.0:
success = False
if success:
total_success += 1
l2_total += l2_distortion
suffix = "id{}_seq{}_prev{}_adv{}_{}_dist{}".format(all_true_ids[i], i, original_class[-1], adversarial_class[-1], success, l2_distortion)
print("Saving to", suffix)
show(inputs, "{}/{}/{}_original_{}.png".format(args['save'], args['dataset'], img_no, suffix))
show(adv, "{}/{}/{}_adversarial_{}.png".format(args['save'], args['dataset'], img_no, suffix))
show(adv - inputs, "{}/{}/{}_diff_{}.png".format(args['save'], args['dataset'], img_no, suffix))
print("[STATS][L1] total = {}, seq = {}, id = {}, time = {:.3f}, success = {}, const = {:.6f}, prev_class = {}, new_class = {}, distortion = {:.5f}, success_rate = {:.3f}, l2_avg = {:.5f}".format(img_no, i, all_true_ids[i], timeend - timestart, success, const, original_class[-1], adversarial_class[-1], l2_distortion, total_success / float(img_no), 0 if total_success == 0 else l2_total / total_success))
sys.stdout.flush()
# t = np.random.randn(28*28).reshape(1,28,28,1)
# print(model.model.predict(t))
if __name__ == "__main__":
import argparse
parser = argparse.ArgumentParser()
parser.add_argument("-d", "--dataset", choices=["mnist", "cifar10", "imagenet"], default="mnist")
parser.add_argument("-s", "--save", default="./saved_results")
parser.add_argument("-a", "--attack", choices=["white", "black"], default="white")
parser.add_argument("-n", "--numimg", type=int, default=0, help = "number of test images to attack")
parser.add_argument("-m", "--maxiter", type=int, default=0, help = "set 0 to use default value")
parser.add_argument("-p", "--print_every", type=int, default=100, help = "print objs every PRINT_EVERY iterations")
parser.add_argument("-o", "--early_stop_iters", type=int, default=100, help = "print objs every EARLY_STOP_ITER iterations, 0 is maxiter//10")
parser.add_argument("-f", "--firstimg", type=int, default=0)
parser.add_argument("-b", "--binary_steps", type=int, default=0)
parser.add_argument("-c", "--init_const", type=float, default=0.0)
parser.add_argument("-z", "--use_zvalue", action='store_true')
parser.add_argument("-u", "--untargeted", action='store_true')
parser.add_argument("-r", "--reset_adam", action='store_true', help = "reset adam after an initial solution is found")
parser.add_argument("--use_resize", action='store_true', help = "resize image (only works on imagenet!)")
parser.add_argument("--adam_beta1", type=float, default=0.9)
parser.add_argument("--adam_beta2", type=float, default=0.999)
parser.add_argument("--seed", type=int, default=1216)
parser.add_argument("--solver", choices=["adam", "newton", "adam_newton", "fake_zero"], default="adam")
parser.add_argument("--save_ckpts", default="", help = "path to save checkpoint file")
parser.add_argument("--load_ckpt", default="", help = "path to numpy checkpoint file")
parser.add_argument("--start_iter", default=0, type=int, help = "iteration number for start, useful when loading a checkpoint")
parser.add_argument("--init_size", default=32, type=int, help = "starting with this size when --use_resize")
parser.add_argument("--uniform", action='store_true', help = "disable importance sampling")
args = vars(parser.parse_args())
# add some additional parameters
# learning rate
args['lr'] = 1e-2
args['inception'] = False
args['use_tanh'] = True
# args['use_resize'] = False
if args['maxiter'] == 0:
if args['attack'] == "white":
args['maxiter'] = 1000
else:
if args['dataset'] == "imagenet":
if args['untargeted']:
args['maxiter'] = 1500
else:
args['maxiter'] = 50000
elif args['dataset'] == "mnist":
args['maxiter'] = 3000
else:
args['maxiter'] = 1000
if args['init_const'] == 0.0:
if args['binary_steps'] != 0:
args['init_const'] = 0.01
else:
args['init_const'] = 0.5
if args['binary_steps'] == 0:
args['binary_steps'] = 1
# set up some parameters based on datasets
if args['dataset'] == "imagenet":
args['inception'] = True
args['lr'] = 2e-3
# args['use_resize'] = True
# args['save_ckpts'] = True
# for mnist, using tanh causes gradient to vanish
if args['dataset'] == "mnist":
args['use_tanh'] = False
# when init_const is not specified, use a reasonable default
if args['init_const'] == 0.0:
if args['binary_search']:
args['init_const'] = 0.01
else:
args['init_const'] = 0.5
# setup random seed
random.seed(args['seed'])
np.random.seed(args['seed'])
print(args)
main(args)