-
Notifications
You must be signed in to change notification settings - Fork 15
/
smoothing_evaluation.py
164 lines (139 loc) · 6.41 KB
/
smoothing_evaluation.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
"""
Run randomized certification on the test set. Loosely based on code from
https://github.com/locuslab/smoothing
"""
import argparse
import os
from time import time
import datetime
from utils import get_model
import logging
import pandas as pd
import torch
import torch.nn
import torch.nn.functional as F
from datasets import SemiSupervisedDataset, DATASETS
from torchvision import datasets, transforms
from smoothing import Smooth
import numpy as np
if __name__ == "__main__":
parser = argparse.ArgumentParser(
description='Randomized smoothing certification')
parser.add_argument('--dataset', type=str, default='cifar10',
choices=DATASETS,
help='The dataset')
parser.add_argument('--data_dir', default='data', type=str,
help='Directory where datasets are located')
parser.add_argument('--model_path',
help='Model for certification')
parser.add_argument('--model', '-m', default='wrn-28-10', type=str,
help='Name of the model')
parser.add_argument('--output_name', default='smoothing', type=str,
help='Name of output files')
parser.add_argument("--sigma", default=0.25, type=float,
help="Noise hyperparameter")
parser.add_argument("--batch", type=int, default=1000, help="Batch size")
parser.add_argument("--skip", type=int, default=1,
help="Skip this many examples between each examples "
"we certify")
parser.add_argument("--max", type=int, default=-1,
help="Stop when example index == max")
# parser.add_argument("--split", choices=["train", "test"], default="test",
# help="train or test set")
parser.add_argument("--N0", type=int, default=100,
help="Number of noise sample for classification "
"decision")
parser.add_argument("--N", type=int, default=10000,
help="Number of noise sample for radius evaluation")
parser.add_argument("--alpha", type=float, default=0.001,
help="Failure probability")
parser.add_argument('--random_seed', type=int, default=None)
args = parser.parse_args()
# create log
output_dir, _ = os.path.split(args.model_path)
logging.basicConfig(
level=logging.INFO,
format="%(asctime)s | %(message)s",
handlers=[
logging.FileHandler(os.path.join(output_dir,
'%s.log' % args.output_name)),
logging.StreamHandler()
])
logger = logging.getLogger()
logging.info('Randomized smoothing certification')
logging.info('Args: %s', args)
# also store stuff in data frame
df = pd.DataFrame()
# load the base classifier
checkpoint = torch.load(args.model_path)
state_dict = checkpoint.get('state_dict', checkpoint)
if not all(k.startswith('module') for k in state_dict):
state_dict = {'module.' + k: v for k, v in state_dict.items()}
num_classes = checkpoint.get('num_classes', 10)
normalize_input = checkpoint.get('normalize_input', False)
base_classifier = get_model(args.model, num_classes=num_classes,
normalize_input=normalize_input)
base_classifier = torch.nn.DataParallel(base_classifier).cuda()
# setting loader to be non-strict so we can load Cohen et al.'s model
base_classifier.load_state_dict(state_dict,
strict=(args.model != 'resnet-110'))
# create the smooothed classifier g
smoothed_classifier = Smooth(base_classifier, num_classes, args.sigma)
# iterate through the dataset
transform_test = transforms.ToTensor()
# dataset = datasets.CIFAR10(root='data', train=False,
# download=True,
# transform=transform_test)
dataset = SemiSupervisedDataset(base_dataset=args.dataset,
train=False, root=args.data_dir,
download=True,
transform=transform_test)
# Shuffling the dataset if random seed is not None
if args.random_seed is not None:
np.random.seed(args.random_seed)
np.random.shuffle(dataset.targets)
np.random.seed(args.random_seed)
np.random.shuffle(dataset.data)
filename = args.output_name + '_seed_' + str(args.random_seed) + '.csv'
else:
filename = args.output_name + '.csv'
if os.path.exists(os.path.join(output_dir, filename)):
logging.info('Output file exists, resuming...')
df = pd.read_csv(os.path.join(output_dir, filename), index_col=0)
i_start = int(df.i.values[-1]) + 1
is_correct = list(df.correct.values)
is_rob_correct = list(df.correct.values *
(df.radius.values >= args.sigma))
else:
i_start = 0
is_correct = []
is_rob_correct = []
for i in range(i_start, len(dataset)):
# only certify every args.skip examples, and stop
# after args.max examples
if i % args.skip != 0:
continue
if i == args.max:
break
(x, label) = dataset[i]
before_time = time()
# certify the prediction of g around x
x = x.cuda()
prediction, pAbar, radius, counts = smoothed_classifier.certify(
x, args.N0, args.N, args.alpha, args.batch)
after_time = time()
correct = int(prediction == label)
is_correct.append(correct)
is_rob_correct.append(correct * (radius >= args.sigma))
time_elapsed = str(
datetime.timedelta(seconds=(after_time - before_time)))
logging.info("{}/{} | correct={}, pAbar={:.5f}, radius={:.3f}, "
"clean accuracy={:.1f}%, robust accuracy={:.1f}%".format(
i + 1, len(dataset), correct, pAbar, radius,
100 * np.mean(is_correct), 100 * np.mean(is_rob_correct)))
df = df.append(pd.Series(dict(i=i, label=label, prediction=prediction,
pAbar=pAbar, radius=radius,
correct=correct, counts=counts,
time_elapsed=time_elapsed)),
ignore_index=True)
df.to_csv(os.path.join(output_dir, filename))