def attack_transform_overlap(
attack_name,
transform_name,
transforms,
reduce_mode="none",
result_dir="result/lenet/9transform_attack_overlap"):
name = attack_name + '_' + transform_name
lenet_mnist_class_trace = class_trace(
trace_name,
model_config=LENET,
data_config=data_config,
)
threshold = 0.5
# DeepFool will shutdown when num_gpu<0.2
num_gpus = 0.2
overlap_fn = calc_trace_side_overlap
per_channel = False
path = os.path.join(result_dir, f"{name}_overlap.csv")
# print(f"Computing {name}")
# lenet_overlap_ratio = attack_reduced_edge_detection_count_violation(
lenet_overlap_ratio = attack_reduced_edge_detection(
attack_name=attack_name,
attack_fn=attacks[attack_name][0],
generate_adversarial_fn=cw_generate_adversarial_example if
attack_name.startswith("CW") else foolbox_generate_adversarial_example,
class_trace_fn=lambda class_id: lenet_mnist_class_trace(
class_id,
threshold,
label=model_label,
trace_dir=trace_dir,
),
select_fn=lambda input: arg_approx(input, threshold),
overlap_fn=overlap_fn,
path=path,
per_channel=per_channel,
preprocessing=(0.1307, 0.3081),
image_size=28,
class_num=10,
norm_fn=mnist.normalize,
data_format="channels_first",
**(attacks[attack_name][1] if len(attacks[attack_name]) == 2 else {}),
images_per_class=images_per_class,
model_dir=model_dir,
num_gpus=num_gpus,
transforms=transforms,
transform_name=transform_name,
reduce_mode=reduce_mode,
)
lenet_overlap_ratio.save()
return lenet_overlap_ratio.load()
def visualize_features(
transforms,
result_dir = "result/lenet/model_augmentation/visualize"
):
name = attack_name
lenet_mnist_class_trace = class_trace(
trace_name,
model_config=LENET,
data_config=data_config,
)
threshold = 0.5
# DeepFool will shutdown when num_gpu<0.2
num_gpus = 0.2
overlap_fn = calc_trace_side_overlap
per_channel = False
path = os.path.join(result_dir, f"{name}_overlap.csv")
print(f"Computing {name}")
lenet_overlap_ratio = test_data_forward(
class_trace_fn=lambda class_id: lenet_mnist_class_trace(
class_id,
threshold,
label=model_label,
trace_dir = trace_dir,
),
select_fn=lambda input: arg_approx(input, threshold),
overlap_fn=overlap_fn,
path=path,
per_channel=per_channel,
preprocessing=(0.1307, 0.3081),
image_size=28,
class_num=10,
norm_fn=mnist.normalize,
data_format="channels_first",
**(attacks[attack_name][1] if len(attacks[attack_name]) == 2 else {}),
images_per_class=images_per_class,
model_dir=model_dir,
num_gpus = num_gpus,
transforms = transforms,
)
# "CWL2": [CarliniL2],
}
adversarial_label = 1
normal_label = -1
class_trace_fn = lambda class_id: lenet_mnist_class_trace(
class_id,
threshold,
label=model_label,
trace_dir=trace_dir,
)
lenet_mnist_class_trace = class_trace(
trace_name,
model_config=LENET,
data_config=data_config,
)
def filter_edge_by_key(
trace: AttrMap,
key: str = TraceKey.EDGE,
graph=LENET.network_class.graph().load(),
):
reconstruct_edge_fn = partial(
reconstruct_edge,
trace,
graph,
key,
)
# Configs
result_dir = "result/resnet18cifar10"
# The label of LeNet model
model_label = "dropout"
model_dir = f"{result_dir}/model_{model_label}"
# Dir to save traces for this model
trace_dir = f"{model_dir}/traces_{threshold}"
# The name of traces for this experiment.
# The final dir to save traces is trace_dir/trace_name.
# Should adapt with transform_list
trace_name = "noop"
# Example number per class to extract trace
example_num = 1
resnet_18_cifar10_class_trace = class_trace(name=trace_name,
model_config=RESNET_18_CIFAR10,
data_config=CIFAR10_TRAIN)
# resnet_18_cifar10_class_trace_growth = class_trace_growth(
# name=name, model_config=RESNET_18_CIFAR10, data_config=CIFAR10_TRAIN
# )
resnet_18_cifar10_class_trace_compact = class_trace_compact(
resnet_18_cifar10_class_trace,
name=trace_name,
model_config=RESNET_18_CIFAR10)
# save_resnet_18_cifar10_class_traces_low_latency = save_class_traces_low_latency(
# name=name, model_config=RESNET_18_CIFAR10, data_config=CIFAR10_TRAIN
# )