Skip to content

Update universal_pert.py #6

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ANONYMOUS-GURU wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Update universal_pert.py #6

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
63 changes: 40 additions & 23 deletions python/universal_pert.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,9 +40,14 @@ def universal_perturbation(dataset, f, grads, delta=0.2, max_iter_uni = np.inf,

:return: the universal perturbation.
"""

v = 0
v=0
fooling_rate = 0.0
import os
file_perturbation=os.path.join(os.getcwd(),'data','universal_partial.npy')
if os.path.isfile(file_perturbation) == 1:
v = np.load(file_perturbation)
fooling_rate=fooling_rate_calc(v,dataset,f)
print('>> Loaded perturbation file from disk perturbation fooling_rate obesrved :: ',fooling_rate)
num_images = np.shape(dataset)[0] # The images should be stacked ALONG FIRST DIMENSION

itr = 0
Expand All @@ -65,30 +70,42 @@ def universal_perturbation(dataset, f, grads, delta=0.2, max_iter_uni = np.inf,
# Make sure it converged...
if iter < max_iter_df-1:
v = v + dr

# Project on l_p ball
v = proj_lp(v, xi, p)
print('>> iters required ::',iter)
else:
print('>> k = ', k, ', pass #', itr,'this k to deepfool no convergence for 10 iters')

itr = itr + 1

# Perturb the dataset with computed perturbation
dataset_perturbed = dataset + v

est_labels_orig = np.zeros((num_images))
est_labels_pert = np.zeros((num_images))
else:
print('>> k = ', k, ', pass #', itr,'Perturb success for this image without deepfool')

batch_size = 100
num_batches = np.int(np.ceil(np.float(num_images) / np.float(batch_size)))

# Compute the estimated labels in batches
for ii in range(0, num_batches):
m = (ii * batch_size)
M = min((ii+1)*batch_size, num_images)
est_labels_orig[m:M] = np.argmax(f(dataset[m:M, :, :, :]), axis=1).flatten()
est_labels_pert[m:M] = np.argmax(f(dataset_perturbed[m:M, :, :, :]), axis=1).flatten()

# Compute the fooling rate
fooling_rate = float(np.sum(est_labels_pert != est_labels_orig) / float(num_images))
itr = itr + 1
np.save(file_perturbation,v)
fooling_rate=fooling_rate_calc(v,dataset,f)
print('FOOLING RATE = ', fooling_rate)

return v

# Perturb the dataset with computed perturbation
def fooling_rate_calc(v,dataset,f):
dataset_perturbed = dataset + v
num_images = np.shape(dataset)[0]
est_labels_orig = np.zeros((num_images))
est_labels_pert = np.zeros((num_images))

batch_size = 100
num_batches = np.int(np.ceil(np.float(num_images) / np.float(batch_size)))

# Compute the estimated labels in batches
for ii in range(0, num_batches):
m = (ii * batch_size)
M = min((ii+1)*batch_size, num_images)
est_labels_orig[m:M] = np.argmax(f(dataset[m:M, :, :, :]), axis=1).flatten()
est_labels_pert[m:M] = np.argmax(f(dataset_perturbed[m:M, :, :, :]), axis=1).flatten()

# Compute the fooling rate
fooling_rate = float(np.sum(est_labels_pert != est_labels_orig) / float(num_images))
return fooling_rate