WeMeM/datasets.py at main · CactiLab/WeMeM · GitHub

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
import os
import pickle
import numpy as np
import pandas as pd
import sklearn
import torch
import torchvision
from sklearn.model_selection import train_test_split
import torchvision.transforms as transforms
from torch.utils.data import TensorDataset, Subset


def get_dataset(name, train=True):
    print(f"Build Dataset {name}")
    if name == "cifar10":
        mean = (0.4914, 0.4822, 0.4465)
        std = (0.2023, 0.1994, 0.2010)
        transform = transforms.Compose([
            transforms.ToTensor(),
            transforms.Normalize(mean, std),
        ])
        dataset = torchvision.datasets.CIFAR10(root='./data/datasets/cifar10-data', train=train, download=True, transform=transform)
    elif name == "cifar100":
        mean = (0.5071, 0.4867, 0.4408)
        std = (0.2675, 0.2565, 0.2761)
        transform = transforms.Compose([
            transforms.ToTensor(),
            transforms.Normalize(mean, std),
        ])
        dataset = torchvision.datasets.CIFAR100(root='./data/datasets/cifar100-data', train=train, download=True, transform=transform)
    elif name == "mnist":
        mean = (0.1307,)
        std = (0.3081,)
        transform = transforms.Compose([transforms.ToTensor(),
                                        transforms.Normalize(mean, std)
                                        ])
        dataset = torchvision.datasets.MNIST(root='data/datasets/mnist-data', train=train, download=True,
                                             transform=transform)
    elif name == "cinic":
        # the dataset can be downloaded from https://datashare.ed.ac.uk/bitstream/handle/10283/3192/CINIC-10.tar.gz?sequence=4&isAllowed=y
        if not os.path.exists("./data/datasets/cinic/cinic.pkl"):
            cinic_directory = './data/datasets/cinic'
            cinic_mean = [0.47889522, 0.47227842, 0.43047404]
            cinic_std = [0.24205776, 0.23828046, 0.25874835]
            transform = transforms.Compose([
                transforms.ToTensor(),
                transforms.Normalize(mean=cinic_mean,std=cinic_std),
            ])
            trainset = torchvision.datasets.ImageFolder(cinic_directory + '/train', transform=transform)
            testset = torchvision.datasets.ImageFolder(cinic_directory + '/test', transform=transform)

            with open("./data/datasets/cinic/train_test_idx.pkl", 'rb') as f:
                trainidx, testidx= pickle.load(f)
            train_data = Subset(trainset, trainidx)
            test_data = Subset(testset, testidx)
            with open("./data/datasets/cinic/cinic.pkl", "wb") as f:
                pickle.dump([train_data, test_data], f)
        else:
            with open("./data/datasets/cinic/cinic.pkl", "rb") as f:
                train_data, test_data = pickle.load(f)
        if train == False:
            dataset = test_data
        else:
            dataset = train_data

    elif name == "texas100":
        # the dataset can be downloaded from https://www.comp.nus.edu.sg/~reza/files/dataset_texas.tgz
        if not os.path.exists("./data/datasets/texas/texas100.pkl"):
            x = np.loadtxt("./data/datasets/texas/feats.txt", delimiter=',')
            x_data = torch.tensor(x[:, :]).float()
            y = np.loadtxt("./data/datasets/texas/labels.txt", delimiter=',')
            y_data = torch.tensor(y[:] - 1).long()
            dataset = TensorDataset(x_data, y_data)
            trainset, testset = train_test_split(list(range(len(dataset))), test_size=0.2) # Make sure to calculate the mem-score for these training data.
            train_dataset = Subset(dataset, trainset)
            test_dataset = Subset(dataset, testset)
            with open("./data/datasets/texas/texas100.pkl", 'wb') as f:
                pickle.dump([train_dataset, test_dataset], f)
        else:
            with open("./data/datasets/texas/texas100.pkl", 'rb') as f:
                train_dataset, test_dataset = pickle.load(f)
        if train == False:
            dataset = test_dataset
        else:
            dataset = train_dataset

    elif name == "location":
        # the dataset can be downloaded from https://github.com/jjy1994/MemGuard/tree/master/data/location
        if not os.path.exists("./data/datasets/location/location.pkl"):
            dataset = np.load("./data/datasets/location/data_complete.npz")
            x_data = torch.tensor(dataset['x'][:, :]).float()
            y_data = torch.tensor(dataset['y'][:] - 1).long()
            dataset = TensorDataset(x_data, y_data)
            trainset, testset = train_test_split(list(range(len(dataset))), test_size=0.2) # Make sure to calculate the mem-score for these training data.
            train_dataset = Subset(dataset, trainset)
            test_dataset = Subset(dataset, testset)
            with open("./data/datasets/location/location.pkl", 'wb') as f:
                pickle.dump([train_dataset, test_dataset], f)
        else:
            with open("./data/datasets/location/location.pkl", 'rb') as f:
                train_dataset, test_dataset = pickle.load(f)
        if train == False:
            dataset = test_dataset
        else:
            dataset = train_dataset

    elif name == "purchase100":
        # the dataset can be downloaded from https://www.comp.nus.edu.sg/~reza/files/dataset_purchase.tgz
        if not os.path.exists("./data/datasets/purchase100/purchase100.pkl"):
            dataset = np.loadtxt("./data/datasets/purchase100/purchase100.txt", delimiter=',')
            x_data = torch.tensor(dataset[:, 1:], dtype=torch.float32)
            y_data = torch.tensor(dataset[:, 0] - 1, dtype=torch.long)
            dataset = TensorDataset(x_data, y_data)
            trainset, testset = train_test_split(list(range(len(dataset))), test_size=0.2) # Make sure to calculate the mem-score for these training data.
            train_dataset = Subset(dataset, trainset)
            test_dataset = Subset(dataset, testset)
            with open("./data/datasets/purchase100/purchase100.pkl", 'wb') as f:
                pickle.dump([train_dataset, test_dataset], f)
        else:
            with open("./data/datasets/purchase100/purchase100.pkl", 'rb') as f:
                train_dataset, test_dataset = pickle.load(f)
        if train == False:
            dataset = test_dataset
        else:
            dataset = train_dataset
    else:
        raise ValueError

    return dataset