COCO/MLP_models.py at main · ZJU-DAILY/COCO · 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
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
import torch
import torch.nn as nn
import numpy as np
from torch.nn.utils.rnn import pad_sequence
from torch.utils.data import Dataset


class MLP(nn.Module):
    def __init__(self, input_dim, max_seq_length, hidden_dim):
        super(MLP, self).__init__()

        self.sequence_fc = nn.Linear(input_dim * max_seq_length, hidden_dim)
        self.additional_fc = nn.Linear(1, hidden_dim)

        self.relu = nn.ReLU()

        self.fusion_fc = nn.Linear(hidden_dim * 2, hidden_dim)
        self.output_fc = nn.Linear(hidden_dim, 1)

    def forward(self, x, additional_feature):

        first_col = x[:, :, 0]
        second_col = x[:, :, 1]
        sorted_indices = torch.argsort(-first_col * 1e5 - second_col, dim=1)
        b, n, c = x.shape
        x = torch.gather(x, 1, sorted_indices.unsqueeze(2).expand(b, n, c))

        x = x.view(x.size(0), -1)
        x = self.sequence_fc(x)
        x = self.relu(x)

        if additional_feature.dim() == 1:
            additional_feature = additional_feature.unsqueeze(1)
        additional_feature = self.additional_fc(additional_feature)
        additional_feature = self.relu(additional_feature)

        combined = torch.cat((x, additional_feature), dim=1)
        combined = self.fusion_fc(combined)
        combined = self.relu(combined)

        output = self.output_fc(combined)
        return output.squeeze()


class MLPNonleaf(nn.Module):
    def __init__(self, input_dim, max_seq_length1, max_seq_length2, hidden_dim):
        super(MLPNonleaf, self).__init__()

        self.sequence_fc1 = nn.Linear(input_dim * max_seq_length1, hidden_dim)
        self.sequence_fc2 = nn.Linear(input_dim * max_seq_length2, hidden_dim)

        self.single_feature_fc = nn.Linear(1, hidden_dim)

        self.relu = nn.ReLU()

        self.fusion_fc = nn.Linear(hidden_dim * 3, hidden_dim)
        self.output_fc = nn.Linear(hidden_dim, 1)

    # @profile
    def forward(self, sequence1, sequence2, single_feature):

        first_col = sequence1[:, :, 0]
        second_col = sequence1[:, :, 1]
        sorted_indices = torch.argsort(-first_col * 1e5 - second_col, dim=1)
        b, n, c = sequence1.shape
        sequence1 = torch.gather(
            sequence1, 1, sorted_indices.unsqueeze(2).expand(b, n, c)
        )

        x1 = sequence1.view(sequence1.size(0), -1)
        x1 = self.sequence_fc1(x1)
        x1 = self.relu(x1)

        first_col = sequence2[:, :, 0]
        second_col = sequence2[:, :, 1]
        sorted_indices = torch.argsort(-first_col * 1e5 - second_col, dim=1)
        b, n, c = sequence2.shape
        sequence2 = torch.gather(
            sequence2, 1, sorted_indices.unsqueeze(2).expand(b, n, c)
        )

        x2 = sequence2.view(sequence2.size(0), -1)
        x2 = self.sequence_fc2(x2)
        x2 = self.relu(x2)

        if single_feature.dim() == 1:
            single_feature = single_feature.unsqueeze(1)
        x3 = self.single_feature_fc(single_feature)
        x3 = self.relu(x3)

        combined = torch.cat((x1, x2, x3), dim=1)
        combined = self.fusion_fc(combined)
        combined = self.relu(combined)

        output = self.output_fc(combined)
        return output.squeeze()


def preprocess_sequences(sequences, max_length):
    sequences = [
        torch.tensor(seq[:max_length], dtype=torch.float32) for seq in sequences
    ]
    sequences_padded = pad_sequence(sequences, batch_first=True, padding_value=0)
    if sequences_padded.size(1) < max_length:
        pad_size = max_length - sequences_padded.size(1)
        padding = torch.zeros(
            (sequences_padded.size(0), pad_size, sequences_padded.size(2)),
            dtype=torch.float32,
        )
        sequences_padded = torch.cat((sequences_padded, padding), dim=1)

    return sequences_padded


class SequenceDataset(Dataset):
    def __init__(self, sequences, labels, additional_features, max_seq_length):
        self.sequences = preprocess_sequences(sequences, max_seq_length)
        self.labels = torch.tensor(labels, dtype=torch.float32)
        self.additional_features = torch.tensor(
            additional_features, dtype=torch.float32
        )

    def __len__(self):
        return len(self.labels)

    def __getitem__(self, idx):
        return self.sequences[idx], self.additional_features[idx], self.labels[idx]


class SequenceDatasetNonleaf(Dataset):
    def __init__(
        self,
        sequences,
        sequences_range,
        labels,
        additional_features,
        max_seq_length1,
        max_seq_length2,
    ):
        self.sequences = preprocess_sequences(sequences, max_seq_length1)
        self.sequences_range = preprocess_sequences(
            sequences_range, max_seq_length2
        )  # New
        self.labels = torch.tensor(labels, dtype=torch.float32)
        self.additional_features = torch.tensor(
            additional_features, dtype=torch.float32
        )

    def __len__(self):
        return len(self.labels)

    def __getitem__(self, idx):
        return (
            self.sequences[idx],
            self.sequences_range[idx],
            self.additional_features[idx],
            self.labels[idx],
        )