-
Notifications
You must be signed in to change notification settings - Fork 1
/
gcn.py
266 lines (225 loc) · 8.91 KB
/
gcn.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
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
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
import random
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch_geometric.data import Data
from torch_geometric.nn import GATConv, GCNConv
from torch_geometric.datasets import Planetoid
from torch_geometric.datasets import CitationFull
from torch_geometric.datasets import Coauthor
from torch_geometric.datasets import AMiner
from torch_geometric.datasets import Amazon
import torch_geometric.transforms as T
import scipy.sparse as sp
import networkx as nx
import csv
# from utils import MADGap
from utils import load_CoraFull, load_Coauthor, load_Amazon, encode_onehot, get_train_val_test_split, normalize_adj,\
sample_per_class
# 'Cora' or 'CiteSeer'
name = 'AmazonPhoto'
# 'MADGap' or 'Accuracy'
evaluate_name = 'Accuracy'
torch.manual_seed(42)
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
torch.cuda.set_device(1)
def load_dataset(root='./data/',name='AmazonCS', seed=42):
print('loading data {} from {}'.format(name, root))
if name == 'CoraFull':
dataset = load_CoraFull()
elif name == 'CoauthorCS':
dataset = load_Coauthor(name='CS')
elif name == 'AmazonComputers':
dataset = load_Amazon(name='Computers')
elif name == 'AmazonPhoto':
dataset = load_Amazon(name='Photo')
else:
print('error dataset name input')
edge_index = dataset.data.edge_index
labels = dataset.data.y
labels_onehot = encode_onehot(labels.numpy())
print(labels_onehot.shape)
num_samples, num_classes = labels_onehot.shape
if name == 'CoraFull': # ignore the class whose node number less than 50
deleted_class = []
deleted_nodes = []
for class_index in [68, 69]:
nodes = []
for sample_index in range(num_samples):
if labels_onehot[sample_index, class_index] > 0.0:
nodes.append(sample_index)
if len(nodes) < 50:
print('ignore class index={}, ignore node num={}'.format(class_index, len(nodes)))
deleted_class.append(class_index)
deleted_nodes.extend(nodes)
adj = sp.coo_matrix((torch.ones_like(edge_index[0]), edge_index), shape=(num_samples, num_samples))
adj = adj + adj.T.multiply(adj.T > adj) - adj.multiply(adj.T > adj)
adj = np.array(adj.todense())
adj = np.delete(adj, deleted_nodes, axis=0)
adj = np.delete(adj, deleted_nodes, axis=1)
new_edge_index = nx.to_edgelist(nx.from_numpy_matrix(adj))
new_edge_index = torch.LongTensor(list(map(lambda x: x[:2], new_edge_index))).t()
new_edge_index_b = torch.vstack((new_edge_index[1],new_edge_index[0]))
new_edge_index = torch.cat((new_edge_index, new_edge_index_b), dim=1)
features = np.delete(dataset.data.x, deleted_nodes, axis=0)
labels = np.delete(labels, deleted_nodes, axis=0)
labels_onehot = encode_onehot(labels.numpy())
dataset.data.edge_index = new_edge_index
dataset.data.y = torch.LongTensor(labels)
dataset.data.x = torch.FloatTensor(features)
print(dataset.data.edge_index.shape)
print(dataset.data.x.shape)
print(dataset.data.y.shape)
labels = dataset.data.y
labels_onehot = encode_onehot(labels.numpy())
random_state = np.random.RandomState(seed=seed)
idx_train, idx_val, idx_test = get_train_val_test_split(random_state,labels_onehot, 20, 30)
idx_train = torch.LongTensor(idx_train)
idx_val = torch.LongTensor(idx_test)
idx_test = torch.LongTensor(idx_test)
dataset.train_mask = idx_train
dataset.test_mask = idx_test
return dataset, idx_train, idx_test
def get_mask(dataset, seed, use_label_rate = False, num_example_per_class=20):
labels = dataset.data.y
labels_onehot = encode_onehot(labels.numpy())
random_state = np.random.RandomState(seed=seed)
idx_train, idx_val, idx_test = get_train_val_test_split(random_state, labels_onehot, 20, 30)
idx_train = torch.LongTensor(idx_train)
idx_val = torch.LongTensor(idx_test)
idx_test = torch.LongTensor(idx_test)
return idx_train, idx_test
dataset = None
# dataset.transform = T.NormalizeFeatures()
def main(order):
# 隐藏层
nhid = 16
nhid_layers = []
for i in range(order-1):
nhid_layers.append(nhid)
class GCN(nn.Module):
def __init__(self):
super(GCN, self).__init__()
self.list = [dataset.num_features] + nhid_layers + [dataset.num_classes]
self.layers = []
# 举例子 self.list = [16, 8, 8, 8, 3]
for i, _ in enumerate(self.list[:-1]):
self.layers.append(GCNConv(self.list[i], self.list[i+1]))
self.layers = ListModule(*self.layers)
def forward(self, data):
features, edges = data.x, data.edge_index
if nhid_layers:
for i, _ in enumerate(self.list[:-2]):
features = F.relu(self.layers[i](features, edges))
if i > 0:
features = F.dropout(features, p=0.5, training=self.training)
features = self.layers[i+1](features, edges)
else:
features = self.layers[0](features, edges)
output = F.log_softmax(features, dim=1)
return output, features
class ListModule(nn.Module):
"""
Abstract list layer class.
"""
def __init__(self, *args):
"""
Module initializing.
"""
super(ListModule, self).__init__()
idx = 0
for module in args:
self.add_module(str(idx), module)
idx += 1
def __getitem__(self, idx):
"""
Getting the indexed layer.
"""
if idx < 0 or idx >= len(self._modules):
raise IndexError('index {} is out of range'.format(idx))
it = iter(self._modules.values())
for i in range(idx):
next(it)
return next(it)
def __iter__(self):
"""
Iterating on the layers.
"""
return iter(self._modules.values())
def __len__(self):
"""
Number of layers.
"""
return len(self._modules)
model = GCN().to(device)
data = dataset[0].to(device)
optimizer = torch.optim.Adam(model.parameters(), lr=0.005, weight_decay=5e-4)
# train
model.train()
acc_best = 0
patience = 100
bad_count = 0
for epoch in range(200):
model.train()
optimizer.zero_grad()
output, features = model(data)
loss = F.nll_loss(output[dataset.train_mask], data.y[dataset.train_mask])
loss.backward()
optimizer.step()
# eval
model.eval()
with torch.no_grad():
# _, pred = model(data).max(dim=1)
output, features = model(data)
pred = output.argmax(1)
correct = float(pred[dataset.test_mask].eq(data.y[dataset.test_mask]).sum().item())
acc = correct / dataset.test_mask.shape[0]
if acc > acc_best:
acc_best = acc
else:
bad_count += 1
if bad_count > patience:
print('early stop')
break
model.train()
model.eval()
output, features = model(data)
pred = output.argmax(1)
# print(pred)
if evaluate_name == 'Accuracy':
correct = float(pred[dataset.test_mask].eq(data.y[dataset.test_mask]).sum().item())
acc = correct / dataset.test_mask.shape[0]
return acc
elif evaluate_name == 'MADGap':
madgap = MADGap(dataset, features.cpu().detach().numpy())
madgap_value = madgap.mad_gap_regularizer()
return madgap_value
else:
print('evaluate_name error')
for name in ['CoraFull']:
print('begin train',name)
out = 0
order=2
max = 0
min = 100
seed = random.randint(0, 100)
dataset,_,_ = load_dataset('./data/', name=name, seed=seed)
dataset.transform = T.NormalizeFeatures()
for _ in range(0, 50):
seed = random.randint(0, 100)
train_mask, test_mask = get_mask(dataset, seed)
dataset.train_mask = train_mask
dataset.test_mask = test_mask
print('training gat seed={} ....'.format(seed))
acc = main(order)
if acc > max:
max = acc
if acc < min:
min = acc
print('seed {}, acc {}'.format(seed, acc))
out += acc
out = out / 50
print('acc:', out)
with open('./' + name + '_gcn_' + evaluate_name + '.csv', 'a+') as f:
f.writelines(str(max) + '\t' +str(min)+'\t'+ str(out) + '\n')