-
Notifications
You must be signed in to change notification settings - Fork 0
/
gala.py
124 lines (93 loc) · 3.92 KB
/
gala.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
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
import torch
from torch import Tensor
from torch.utils.tensorboard import SummaryWriter
from torch import nn
import torch.nn.functional as F
import torch_geometric.nn as pyg_nn
from torch_geometric.utils import negative_sampling
from torch_geometric.datasets import Planetoid
from sklearn.cluster import KMeans, SpectralClustering
from sklearn.metrics import normalized_mutual_info_score, adjusted_rand_score
import gcn_deconv
writer = SummaryWriter(log_dir='runs/GALA')
EPS = 1e-15
lr = 0.0001
epochs = 1000
class Encoder(nn.Module):
def __init__(self, num_features):
super().__init__()
self.conv1 = pyg_nn.GCNConv(num_features, 1600, cached=True)
self.conv2 = pyg_nn.GCNConv(1600, 400, cached=True)
def forward(self, x, edge_index):
x = F.relu(self.conv1(x, edge_index))
x = self.conv2(x, edge_index)
return x
class Decoder(nn.Module):
def __init__(self, num_features):
super().__init__()
self.conv1 = gcn_deconv.GCNDeconv(400, 1600, improved=True, cached=True)
self.conv2 = gcn_deconv.GCNDeconv(1600, num_features, improved=True, cached=True)
def forward(self, x, edge_index):
x = F.relu(self.conv1(x, edge_index))
x = self.conv2(x, edge_index)
return x
class GAE(nn.Module):
def __init__(self, num_features):
super().__init__()
self.encoder = Encoder(num_features)
self.decoder = Decoder(num_features)
GAE.reset_parameters(self)
def reset_parameters(self):
pyg_nn.inits.reset(self.encoder)
pyg_nn.inits.reset(self.decoder)
def forward(self, x, edge_index):
return self.encoder(x, edge_index)
def encode(self, x, edge_index):
return self.encoder(x, edge_index)
def decode(self, x, edge_index):
return self.decoder(x, edge_index)
def adjrec(self, z, edge_index, sigmoid=True):
value = (z[edge_index[0]] * z[edge_index[1]]).sum(dim=1)
return torch.sigmoid(value) if sigmoid else value
def adj_loss(self, z, pos_edge_index, neg_edge_index = None):
pos_loss = -torch.log(self.adjrec(z, pos_edge_index) + EPS).mean()
if neg_edge_index is None:
neg_edge_index = negative_sampling(pos_edge_index, z.size(0))
neg_loss = -torch.log(1 - self.adjrec(z, neg_edge_index, sigmoid=True) + EPS).mean()
return pos_loss + neg_loss
def x_loss(self, x, x_hat):
return torch.square(torch.norm(x - x_hat)) / 2 / x.shape[0]
def loss(self, x, edge_index):
z = self.encode(x, edge_index)
x_hat = self.decode(z, edge_index)
return self.x_loss(x, x_hat) + self.adj_loss(z, edge_index)
# test node clustering
def test_NC(self, z, y):
kmeans = KMeans(n_clusters=7, n_init=20)
y_pred = kmeans.fit_predict(z.detach().cpu().numpy())
y_true = y.detach().cpu().numpy()
return normalized_mutual_info_score(y_true, y_pred), \
adjusted_rand_score(y_true, y_pred)
device = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu')
dataset = Planetoid(root='', name='Cora')
data = dataset[0].to(device)
model = GAE(dataset.num_features).to(device)
print(model)
print('%d parameters' % sum(p.numel() for p in model.parameters() if p.requires_grad))
optimizer = torch.optim.Adam(model.parameters(), lr=lr)
for epoch in range(epochs):
model.train()
optimizer.zero_grad()
loss_value = model.loss(data.x, data.edge_index)
loss_value.backward()
optimizer.step()
model.eval()
with torch.no_grad():
z = model.encode(data.x, data.edge_index)
nmi, ari = model.test_NC(z, data.y)
writer.add_scalar("loss", loss_value, global_step=epoch)
writer.add_scalar("nmi", nmi, global_step=epoch)
writer.add_scalar("ari", ari, global_step=epoch)
print(f'Epoch: {epoch:03d}, Loss: {loss_value.float():.4f}, NMI: {nmi:.4f}', f'ARI: {ari:.4f}')
writer.flush()
writer.close()