-
Notifications
You must be signed in to change notification settings - Fork 2
/
label_inference_demo.py
85 lines (66 loc) · 3.04 KB
/
label_inference_demo.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
import sys
import numpy as np
import torch
from torchvision import transforms, datasets
from torchvision.utils import save_image
import unsplit.attacks as unsplit
from unsplit.models import *
from unsplit.util import *
dataset = sys.argv[1]
count = 100
# load datasets and initialize client, server, and clone models
# set the split values so that the client model (second part) has depth of one.
if dataset == 'mnist':
trainset = datasets.MNIST('data/mnist', download=True, train=True, transform=transforms.ToTensor())
testset = datasets.MNIST('data/mnist', download=True, train=False, transform=transforms.ToTensor())
client, server, clone = MnistNet(), MnistNet(), MnistNet()
split_layer = 9
grad_index = 8
elif dataset == 'f_mnist':
trainset = datasets.FashionMNIST('data/f_mnist', download=True, train=True, transform=transforms.ToTensor())
testset = datasets.FashionMNIST('data/f_mnist', download=True, train=False, transform=transforms.ToTensor())
client, server, clone = MnistNet(), MnistNet(), MnistNet()
split_layer = 9
grad_index = 8
elif dataset == 'cifar':
trainset = datasets.CIFAR10('data/cifar', download=True, train=True, transform=transforms.ToTensor())
testset = datasets.CIFAR10('data/cifar', download=True, train=False, transform=transforms.ToTensor())
client, server, clone = CifarNet(), CifarNet(), CifarNet()
split_layer = 16
grad_index = 14
trainloader = torch.utils.data.DataLoader(trainset, shuffle=True, batch_size=64)
testloader = torch.utils.data.DataLoader(testset, shuffle=True)
# -- LABEL INFERENCE ATTACK --
client_opt = torch.optim.Adam(client.parameters(), lr=0.001, amsgrad=True)
server_opt = torch.optim.Adam(server.parameters(), lr=0.001, amsgrad=True)
clone_opt = torch.optim.Adam(clone.parameters(), lr=0.001, amsgrad=True)
criterion = torch.nn.CrossEntropyLoss()
results = []
for idx, (image, label) in enumerate(testloader):
if idx == count:
break
# enumerate possible label values
label_vals = [i * torch.ones(len(label)).long() for i in range(10)]
# obtain gradient values from client
client_opt.zero_grad()
server_opt.zero_grad()
server_out = server(image, end=split_layer)
pred = client(server_out, start=split_layer+1)
loss = criterion(pred, label)
loss.backward(retain_graph=True)
target_grad = [param.grad for param in client.parameters()][grad_index]
# obtain clone model's output
clone_opt.zero_grad()
clone_pred = clone(server_out, start=split_layer+1)
# try out all possible labels and pick the one that produces the closest gradient values
pred_label = unsplit.label_inference(clone_pred, clone, target_grad, label_vals, grad_index)
results.append(label.item() == pred_label.item())
print(f'Label: {label.item()} - Predicted: {pred_label.item()}')
# perform training updates
clone_loss = criterion(clone_pred, pred_label)
clone_loss.backward()
client_opt.step()
clone_opt.step()
server_opt.step()
print('Run complete.')
print(f'Label inference accuracy: {sum(results) / count}')