def label_agreement(
test_data,
substitute_model,
query_victim,
victim_input_shape,
substitute_input_shape,
):
"""Returns the number of agreed upon data points between victim and substitute,
thereby measuring the fidelity of an extraction attack"""
limit = int(len(test_data))
if limit >= 100:
# We limit test data to 100 samples for efficiency
limit = 100
x_data, y_data = process_data(test_data, limit)
substitute_result = get_target(substitute_model, x_data,
substitute_input_shape)
victim_result = query_victim(x_data)
agreed = torch.sum(torch.eq(victim_result, substitute_result)).item()
print(
f"Fidelity: Out of {limit} data points, the models agreed upon {agreed}."
)
return agreed
def query_mnist(input_data):
# PrivacyRaven provides built-in query functions
return get_target(model, input_data, (1, 28, 28, 1))
def query_mnist(input_data):
input_data = torch.from_numpy(input_data)
return get_target(model, input_data)
def query_mnist(input_data):
return get_target(model, input_data)