SamplerGAN/metricsClassifier.py at main · HamidrezaGholamrezaei/SamplerGAN · 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
"""
Classifiers implementation for evaluation process

"""

import torch
import torch.nn as nn
from models.inceptionTime import InceptionBlock


class Flatten(nn.Module):
    def __init__(self, out_features):
        super(Flatten, self).__init__()
        self.output_dim = out_features

    def forward(self, x):
        return x.view(-1, self.output_dim)


class Reshape(nn.Module):
    def __init__(self, out_shape):
        super(Reshape, self).__init__()
        self.out_shape = out_shape

    def forward(self, x):
        return x.view(-1, *self.out_shape)


class Classifier(nn.Module):
    def __init__(self, dataset_name, label_dim, time_step=None):
        """
        Classifiers implementation for evaluation process

        Args:
            dataset_name (str): Name of the dataset.
            label_dim (int): Number of classes in the dataset.
            time_step (int): Number of time steps in the dataset.
        """

        super(Classifier, self).__init__()
        self.dataset_name = dataset_name
        self.label_dim = label_dim
        self.time_step = time_step

        if self.dataset_name == "MNIST":
            self.ClassifierModel = nn.Sequential(
                nn.Conv2d(1, 32, kernel_size=5, stride=1),
                nn.ReLU(),
                nn.Conv2d(32, 64, kernel_size=5, stride=1),
                nn.ReLU(),
                nn.Conv2d(64, 128, kernel_size=5, stride=1),
                nn.ReLU(),
                nn.AdaptiveAvgPool2d(output_size=1),
                Flatten(out_features=128),
                nn.Linear(128, self.label_dim),
            )

        else:
            self.ClassifierModel = nn.Sequential(
                Reshape(out_shape=(1, self.time_step)),
                InceptionBlock(
                    in_channels=1,
                    n_filters=32,
                    kernel_sizes=[5, 11, 23],
                    bottleneck_channels=32,
                    use_residual=True,
                    activation=nn.ReLU()
                ),
                InceptionBlock(
                    in_channels=32 * 4,
                    n_filters=32,
                    kernel_sizes=[5, 11, 23],
                    bottleneck_channels=32,
                    use_residual=True,
                    activation=nn.ReLU()
                ),
                InceptionBlock(
                    in_channels=32 * 4,
                    n_filters=32,
                    kernel_sizes=[5, 11, 23],
                    bottleneck_channels=32,
                    use_residual=True,
                    activation=nn.ReLU()
                ),
                InceptionBlock(
                    in_channels=32 * 4,
                    n_filters=32,
                    kernel_sizes=[5, 11, 23],
                    bottleneck_channels=32,
                    use_residual=True,
                    activation=nn.ReLU()
                ),
                InceptionBlock(
                    in_channels=32 * 4,
                    n_filters=32,
                    kernel_sizes=[5, 11, 23],
                    bottleneck_channels=32,
                    use_residual=True,
                    activation=nn.ReLU()
                ),
                nn.AdaptiveAvgPool1d(output_size=1),
                Flatten(out_features=32 * 4 * 1),
                nn.Linear(in_features=4 * 32 * 1, out_features=self.label_dim)
            )

        if torch.cuda.is_available():
            self.ClassifierModel.cuda()

    def load(self, metric, domain):
        """
        Loading best trained classifier based on IS and FID metrics

        Args:
            metric: Name of the evaluation metric: IS/FID.
            domain: Name of the domain: image/time-series.
        """

        model_path = "./classifiers/best_classifier_{}.torch".format(self.dataset_name)
        state = 'best_state_' if domain == "image" else 'best_inception_state_'
        if metric == "IS":
            self.ClassifierModel.load_state_dict(torch.load(model_path, map_location=torch.device('cpu'))
                                                 [state + 'is'], strict=False)
        elif metric == "FID":
            self.ClassifierModel.load_state_dict(torch.load(model_path, map_location=torch.device('cpu'))
                                                 [state + 'fid'], strict=False)
        return self.ClassifierModel