File size: 5,069 Bytes
3ad8be1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
131
132
133
134
135
136
137
138
139
140
141
142
143
import torch
import torch.nn as nn
import torch.nn.functional as F
from dgl.nn.pytorch.conv import GINConv
from dgl.nn.pytorch.glob import SumPooling, AvgPooling, MaxPooling


class ApplyNodeFunc(nn.Module):
    """Update the node feature hv with MLP, BN and ReLU."""
    def __init__(self, mlp):
        super(ApplyNodeFunc, self).__init__()
        self.mlp = mlp
        self.bn = nn.BatchNorm1d(self.mlp.output_dim)

    def forward(self, h):
        h = self.mlp(h)
        h = self.bn(h)
        h = F.relu(h)
        return h


class MLP(nn.Module):
    """MLP with linear output"""
    def __init__(self, num_layers, input_dim, hidden_dim, output_dim):
        """MLP layers construction
        Paramters
        ---------
        num_layers: int
            The number of linear layers
        input_dim: int
            The dimensionality of input features
        hidden_dim: int
            The dimensionality of hidden units at ALL layers
        output_dim: int
            The number of classes for prediction
        """
        super(MLP, self).__init__()
        self.linear_or_not = True  # default is linear model
        self.num_layers = num_layers
        self.output_dim = output_dim

        if num_layers < 1:
            raise ValueError("number of layers should be positive!")
        elif num_layers == 1:
            # Linear model
            self.linear = nn.Linear(input_dim, output_dim)
        else:
            # Multi-layer model
            self.linear_or_not = False
            self.linears = torch.nn.ModuleList()
            self.batch_norms = torch.nn.ModuleList()

            self.linears.append(nn.Linear(input_dim, hidden_dim))
            for layer in range(num_layers - 2):
                self.linears.append(nn.Linear(hidden_dim, hidden_dim))
            self.linears.append(nn.Linear(hidden_dim, output_dim))

            for layer in range(num_layers - 1):
                self.batch_norms.append(nn.BatchNorm1d((hidden_dim)))

    def forward(self, x):
        if self.linear_or_not:
            # If linear model
            return self.linear(x)
        else:
            # If MLP
            h = x
            for i in range(self.num_layers - 1):
                h = F.relu(self.batch_norms[i](self.linears[i](h)))
            return self.linears[-1](h)


class GIN(nn.Module):
    """GIN model"""
    def __init__(self, input_dim, hidden_dim,num_layers, num_mlp_layers=2,
                 dropout=0.1, learn_eps=False, neighbor_pooling_type='sum',JK='sum'):
        """model parameters setting
        Paramters
        ---------
        num_layers: int
            The number of linear layers in the neural network
        num_mlp_layers: int
            The number of linear layers in mlps
        input_dim: int
            The dimensionality of input features
        hidden_dim: int
            The dimensionality of hidden units at ALL layers
        dropout: float
            dropout ratio on the final linear layer
        learn_eps: boolean
            If True, learn epsilon to distinguish center nodes from neighbors
            If False, aggregate neighbors and center nodes altogether.
        neighbor_pooling_type: str
            how to aggregate neighbors (sum, mean, or max)
        """
        super(GIN, self).__init__()
        self.num_layers = num_layers
        self.learn_eps = learn_eps

        # List of MLPs
        self.ginlayers = torch.nn.ModuleList()
        self.batch_norms = torch.nn.ModuleList()

        for layer in range(self.num_layers - 1):
            if layer == 0:
                mlp = MLP(num_mlp_layers, input_dim, hidden_dim, hidden_dim)
            else:
                mlp = MLP(num_mlp_layers, hidden_dim, hidden_dim, hidden_dim)

            self.ginlayers.append(
                GINConv(ApplyNodeFunc(mlp), neighbor_pooling_type, 0, self.learn_eps))
            self.batch_norms.append(nn.BatchNorm1d(hidden_dim))

        # Linear function for graph poolings of output of each layer
        # which maps the output of different layers into a prediction score

        self.drop = nn.Dropout(dropout)
        self.JK = JK


    def forward(self, g, Perturb=None):
        # list of hidden representation at each layer (including input)
        h = g.ndata.pop('h').float()
        hidden_rep = []
        for i in range(self.num_layers - 1):
            if i == 0 and Perturb is not None:
                h = h + Perturb
            h = self.ginlayers[i](g, h)
            h = self.batch_norms[i](h)
            h = F.relu(h)
            h = self.drop(h)
            hidden_rep.append(h)

        if self.JK=='sum':
            hidden_rep = [h.unsqueeze(0) for h in hidden_rep]
            return torch.sum(torch.cat(hidden_rep, dim=0), dim=0)
        elif self.JK=='max':
            hidden_rep = [h.unsqueeze(0) for h in hidden_rep]
            return torch.max(torch.cat(hidden_rep, dim = 0), dim = 0)[0]
        elif self.JK=='concat':
            return torch.cat(hidden_rep, dim = 1)
        elif self.JK=='last':
            return hidden_rep[-1]