-
Notifications
You must be signed in to change notification settings - Fork 1
/
robustness_check_feature.py
401 lines (343 loc) · 18.2 KB
/
robustness_check_feature.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
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
import torch
import os
import os.path as osp
import GCL.losses as L
import GCL.augmentors as A
import torch.nn.functional as F
import GCL.augmentors.learnable_augs as LeA
import GCL.augmentors.manually_augs as MaA
import logging
import joblib
import argparse
import numpy as np
import networkx as nx
import sys
import time
import warnings
import socket
from torch import nn
from tqdm import tqdm
from torch.optim import Adam
from GCL.eval import get_split, SVMEvaluator
from GCL.models import DualBranchContrast
from torch_geometric.nn import GINConv, global_add_pool
from torch_geometric.data import DataLoader, Dataset
from GCL.models.mymodels import GraphNodeEncoder, GraphEncoder, Predictor
from GCL.utils import compute_infonce, cluster_get, CustomDataLoader, compute_cluster_constrain_loss, k_fold, process_degree, process_topo_eigens, process_xtopo_eigens, save_model_evaluator, \
load_model_evaluator, str_to_bool, add_extra_pos_mask, degree, topo_cluster_labels_get, compute_spectral_topo_loss, compute_spectral_feature_loss, print_memory_usage
from sklearn.metrics import f1_score
from datetime import datetime
from general_data_loader import load_dataset_graphcls, get_split_mask, CombinedDataset
from torch_geometric.utils import scatter, to_networkx, to_dense_adj, get_laplacian
from sklearn.model_selection import KFold
from torch_geometric.utils import coalesce, batched_negative_sampling, to_undirected, dropout_edge
from sklearn.exceptions import ConvergenceWarning
warnings.filterwarnings("ignore", category=ConvergenceWarning)
def get_config():
parser = argparse.ArgumentParser()
parser.add_argument("--use_degree", type=str_to_bool, default=True)
parser.add_argument("--use_spectral_fea", type=str_to_bool, default=True)
parser.add_argument("--use_spectral_topo", type=str_to_bool, default=True)
parser.add_argument("--linear", type=str_to_bool, default=False)
parser.add_argument("--old_version", type=str_to_bool, default=False)
parser.add_argument("--mode", type=str, default="robust_fea", choices=["unsup", "semisup", "robust"])
parser.add_argument("--semi_sup_rate", type=float, default=0.1) # 0.1, or 0.01
parser.add_argument("--dataset_name", type=str, default="PROTEINS")
parser.add_argument("--epoch_select", type=str, default="test_max")
parser.add_argument("--gpu", type=int, default=0)
parser.add_argument("--batch_size", type=int, default=128) # 512
parser.add_argument("--epochs", type=int, default=200) # 200
parser.add_argument("--finetune_epochs", type=int, default=40) # 200
parser.add_argument("--n_clusters", type=int, default=5)
parser.add_argument("--par_id", type=int, default=6)
args = parser.parse_args()
return args
def add_white_noise(X, noise_rate, noise_level=1):
m, n = X.shape
X_with_noise = X.clone()
k = int(m*n*noise_rate)
mask = torch.zeros(m,n, device=X.device)
mask[torch.randint(0, m, (k,), device=X.device), torch.randint(0, n, (k,), device=X.device)] = 1
noise = torch.randn((m,n), device=X.device) * noise_level * mask
X_with_noise = X_with_noise + noise
return X_with_noise
def robust_test(encoder_model, dataloader, args, split, noise_rate=0.04, best_evaluator=None, device='cpu'):
encoder_model.eval()
x = []
y = []
for data, _ in dataloader:
data = data.to(device)
num_nodes = data.batch.size(0)
if args.use_degree:
if data.x is None:
data.x = data.degree
# else:
# data.x = torch.concat([data.x, data.degree], dim=1)
else:
if data.x is None:
data.x = torch.ones((num_nodes, 1), dtype=torch.float32, device=data.batch.device)
new_x = add_white_noise(data.x, noise_rate)
edge_weights = torch.ones((data.edge_index.shape[1], 1), device=device)
z, g = encoder_model(new_x, data.edge_index, data.batch, edge_weights=edge_weights, mode="eval")
x.append(g)
y.append(data.y)
x = torch.cat(x, dim=0)
y = torch.cat(y, dim=0)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
if best_evaluator==None:
# result, best_evaluator = SVMEvaluator(linear=False, setting=setting)(x, y, split)
result, best_evaluator = SVMEvaluator(linear=args.linear, epoch_select=args.epoch_select)(x, y, split)
else:
x_test = x[split['test']].detach().cpu().numpy()
y_test = y[split['test']].detach().cpu().numpy()
test_macro = f1_score(y_test, best_evaluator.predict(x_test), average='macro')
test_micro = f1_score(y_test, best_evaluator.predict(x_test), average='micro')
result = {
'micro_f1': test_micro,
'macro_f1': test_macro,
}
return result, best_evaluator
def main():
hostname = socket.gethostname()
print("Current hostname:", hostname)
EIGEN_VEC_NUM = 4
load_dataset = True
x_eigen_load_dataset = True
not_load_exist_model = False
val_gap=10
save_checkpoint_gap=50
args = get_config()
batch_size = args.batch_size
epochs = args.epochs
dataset_name = args.dataset_name
finetune_epochs = args.finetune_epochs
if torch.cuda.is_available():
device = torch.device(f"cuda:{args.gpu}")
else:
device = torch.device("cpu")
# device = torch.device("cpu")
current_time = datetime.now()
print("Current time:", current_time)
# print settings
print("dataset: ", args.dataset_name, "device: ", device, ", mode: ", args.mode, args.linear)
if not osp.exists('./log'):
os.mkdir('./log')
if not osp.exists(f'./log/{dataset_name}'):
os.mkdir(f'./log/{dataset_name}')
logging_path = "./log/{}/results_{}.log".format(dataset_name, args.mode)
logging.basicConfig(filename=logging_path, level=logging.DEBUG, format='%(asctime)s %(message)s')
if not osp.exists('./checkpoints'):
os.mkdir('./checkpoints')
if not osp.exists(f'./checkpoints/{dataset_name}'):
os.mkdir(f'./checkpoints/{dataset_name}')
# data load
path = osp.join(osp.expanduser('~'), 'datasets')
dataset, split = load_dataset_graphcls(path, dataset_name, args)
# description
if isinstance(dataset, Dataset):
# description
data_summary = dataset.get_summary()
num_nodes = dataset.edge_index.max()+1
num_node_features = dataset.num_node_features
num_classes = dataset.num_classes
y = dataset.y
print(data_summary)
print(dataset_name, "num_nodes: ", num_nodes, "num_node_features: ", num_node_features, "num_classes: ", num_classes)
else:
num_node_features = dataset[0].x.shape[1]
num_classes = dataset[0].y.shape[0]
y = torch.asarray([data.y for data in dataset])
print(dataset_name, "num_node_features: ", num_node_features, "num_classes: ", num_classes)
# degree calculating
whole_processed_dataset_save_path = osp.join(path, dataset_name, "whole_processed_dataset.pt")
if osp.exists(whole_processed_dataset_save_path) and load_dataset:
dataset_dict = torch.load(whole_processed_dataset_save_path)
else:
all_degrees = []
all_edge_weights = []
all_graph_id = []
all_num_components = []
all_eigen_values = [] # all non-zero eigen values
all_eigen_vectors = [] # and corresponded eigen vectors
with tqdm(total=len(dataset), desc='(PRE)') as pbar:
for i, data in enumerate(dataset):
node_degree = process_degree(data)
all_degrees.append(node_degree)
pbar.update()
edge_weights = torch.ones((data.edge_index.shape[1],))
all_edge_weights.append(edge_weights)
all_graph_id.append(i)
# combine dataset
dataset_dict = {"dataset": dataset,
"graph_id": all_graph_id,
"train_mask": torch.zeros(len(dataset), dtype=bool),
"degree": all_degrees,
"edge_weights":all_edge_weights,
}
torch.save(dataset_dict, whole_processed_dataset_save_path)
# eigen info calculate
if not osp.exists(osp.join(path, dataset_name)):
os.mkdir(osp.join(path, dataset_name))
extra_eig_info_dataset_save_path = osp.join(path, dataset_name, "eig_info.pt")
if osp.exists(extra_eig_info_dataset_save_path) and load_dataset:
eig_info_dict = torch.load(extra_eig_info_dataset_save_path)
else:
all_eigen_values = [] # all non-zero eigen values
all_eigen_vectors = [] # and corresponded eigen vectors
with tqdm(total=len(dataset), desc='(Eig)') as pbar:
for i, data in enumerate(dataset):
g = to_networkx(data, to_undirected=True)
num_components = nx.number_connected_components(g)
all_num_components.append(num_components)
del g
eig_value, eig_vector = process_topo_eigens(data, EIGEN_VEC_NUM, num_components)
all_eigen_values.append(eig_value)
all_eigen_vectors.append(eig_vector)
pbar.update()
eig_info_dict = {"eigen_values": all_eigen_values,
"eigen_vectors": all_eigen_vectors,
"num_component": all_num_components,
}
torch.save(eig_info_dict, extra_eig_info_dataset_save_path)
dataset_dict.update({'num_component': eig_info_dict["num_component"]})
dataset_dict.update({'eigen_values': eig_info_dict["eigen_values"]})
dataset_dict.update({'eigen_vectors': eig_info_dict["eigen_vectors"]})
if num_node_features >= 1:
extra_x_eig_info_dataset_save_path = osp.join(path, dataset_name, "x_eig_info.pt")
X_EIGEN_VEC_NUM = min(num_node_features, EIGEN_VEC_NUM)
if osp.exists(extra_x_eig_info_dataset_save_path) and x_eigen_load_dataset:
x_eig_info_dict = torch.load(extra_x_eig_info_dataset_save_path)
else:
# feature x's sigular vector: U, S, V
all_x_eigen_values = [] # all non-zero eigen values
all_x_eigen_vectors_U = []
all_x_eigen_vectors_V = [] # and corresponded eigen vectors
with tqdm(total=len(dataset), desc='(X_Eig)') as pbar:
for i, data in enumerate(dataset):
x_eig_value, x_eig_vector_U, x_eig_vector_V = process_xtopo_eigens(data, X_EIGEN_VEC_NUM)
all_x_eigen_values.append(x_eig_value)
all_x_eigen_vectors_U.append(x_eig_vector_U)
all_x_eigen_vectors_V.append(x_eig_vector_U)
pbar.update()
x_eig_info_dict = {"x_eigen_values": all_x_eigen_values,
"x_eigen_vectors_U": all_x_eigen_vectors_U,
"x_eigen_vectors_V": all_x_eigen_vectors_V,
}
torch.save(x_eig_info_dict, extra_x_eig_info_dataset_save_path)
dataset_dict.update({'x_eigen_values': x_eig_info_dict["x_eigen_values"]})
dataset_dict.update({'x_eigen_vectors_U': x_eig_info_dict["x_eigen_vectors_U"]})
dataset_dict.update({'x_eigen_vectors_V': x_eig_info_dict["x_eigen_vectors_V"]})
if (num_node_features>=1):
dataset_dict_keys = ["dataset", "graph_id", "train_mask", "degree", "edge_weights",
'num_component', "eigen_values", "eigen_vectors",
"x_eigen_values", "x_eigen_vectors_U", "x_eigen_vectors_V"]
else:
dataset_dict_keys = ["dataset", "graph_id", "train_mask", "degree", "edge_weights",
'num_component', "eigen_values", "eigen_vectors"]
dataset_dict = {key: dataset_dict[key] for key in dataset_dict_keys}
combined_dataset = CombinedDataset(**dataset_dict)
dataloader = CustomDataLoader(dataset=combined_dataset, batch_size=batch_size, shuffle=True, has_node_features=(num_node_features>=1))
if args.use_degree:
input_fea_dim = max(num_node_features, 1)
else:
input_fea_dim = max(num_node_features, 1)
# define augmentations
aug1 = A.Identity()
if num_node_features >= 1:
x_eigen_distance_input_dim = X_EIGEN_VEC_NUM
leA_FD = LeA.LearnableFeatureDroppingBySpectral(input_dim=x_eigen_distance_input_dim, hidden_dim=128).to(device)
aug1 = leA_FD
rand_FM = A.FeatureMasking(pf=0.1)
rand_EA = A.EdgeAdding(pe=0.3)
rand_ED = A.EdgeRemoving(pe=0.3)
edge_attr_input_dim = input_fea_dim*2 + dataset_dict['eigen_vectors'][0].shape[1]
leA_ED = LeA.LearnableEdgeDropping(input_dim=edge_attr_input_dim, hidden_dim=128, temp=1.0).to(device) # edge_attr: subg feas, eigen_vecs
leA_EA = LeA.LearnableEdgeAdding(input_dim=edge_attr_input_dim, hidden_dim=128, sample_edges_ratio=0.2).to(device)
leA_EP = LeA.LearnableEdgePerturbation(input_dim_drop=edge_attr_input_dim, input_dim_add=edge_attr_input_dim, hidden_dim=128, sample_edges_ratio=0.2).to(device)
aug2 = leA_EP
augs_type = [type(aug1).__name__, type(aug2).__name__]
aug1_no_spec = aug2_no_spec = False
if augs_type[0] == "LearnableFeatureDroppingBySpectral":
if args.use_spectral_fea==False:
augs_type[0] += "WithoutSpectral"
aug1_no_spec = True
val_gap=epochs//2
elif augs_type[0] in ["LearnableEdgeDropping", "LearnableEdgeAdding", "LearnableEdgePerturbation"]:
if args.use_spectral_topo==False:
augs_type[0] += "WithoutSpectral"
aug1_no_spec = True
val_gap=epochs
if augs_type[1] == "LearnableFeatureDroppingBySpectral":
if args.use_spectral_fea==False:
augs_type[1] += "WithoutSpectral"
aug2_no_spec = True
val_gap=epochs//2
elif augs_type[1] in ["LearnableEdgeDropping", "LearnableEdgeAdding", "LearnableEdgePerturbation"]:
if args.use_spectral_topo==False:
augs_type[1] += "WithoutSpectral"
aug2_no_spec = True
val_gap=epochs
not_load_exist_model = (aug1_no_spec and aug2_no_spec)
if not_load_exist_model:
val_gap=epochs
print(augs_type)
checkpoints_path = "./checkpoints/{}/{}_{}".format(dataset_name, augs_type[0], augs_type[1])
checkpoints_path_2 = "./checkpoints/{}".format(dataset_name) # for some reason, we have backup checkpoints, in case the checkpoints_path doesn't exist
# print_memory_usage()
################################# FineTune #################################
best_acc = 0
print("-"*40+f"Starting"+"-"*40)
test_acc = "nan"
# load encoder model
# if osp.exists(osp.join(checkpoints_path, f"best_encoder_best_0.pt")):
# encoder_model = torch.load(osp.join(checkpoints_path, f"best_encoder_best_0.pt"), map_location=device)
# elif osp.exists(osp.join(checkpoints_path_2, f"best_encoder_0.pt")):
# encoder_model = torch.load(osp.join(checkpoints_path_2, f"best_encoder_0.pt"), map_location=device)
# else:
# raise FileNotFoundError
encoder_model, evaluator, predictor = load_model_evaluator(checkpoints_path, par_id=args.par_id, device=device, old_version=False)
# if encoder_model==None:
# if dataset_name in ["IMDB-BINARY", "COLLAB"]:
# encoder_model, evaluator, predictor = load_model_evaluator(checkpoints_path_2, par_id=args.par_id, device=device, old_version=True)
# print("load_old_checkpoint")
print(type(evaluator))
encoder_model = encoder_model.to(device)
all_results = {}
for noise_rate in [0, 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4, 0.45, 0.5, 0.55, 0.6]:
all_trail_results = {'micro_f1': [],
'macro_f1': []}
for trial in range(5):
fold_all_results = {'micro_f1': [],
'macro_f1': []}
for fold_i, (train_idx, val_idx, test_idx) in enumerate(k_fold(len(dataset), epoch_select=args.epoch_select)):
split = {'train': torch.asarray(train_idx, device=device, dtype=torch.int64),
'valid': torch.asarray(val_idx, device=device, dtype=torch.int64),
'test': torch.asarray(test_idx, device=device, dtype=torch.int64)}
if noise_rate==0:
test_result, best_evaluator = robust_test(encoder_model, dataloader, args, split,
noise_rate=noise_rate,
best_evaluator=None, device=device)
else:
test_result, best_evaluator = robust_test(encoder_model, dataloader, args, split,
noise_rate=noise_rate,
best_evaluator=None, device=device)
fold_all_results['micro_f1'].append(test_result['micro_f1'])
fold_all_results['macro_f1'].append(test_result['macro_f1'])
all_trail_results['micro_f1'].append(np.mean(fold_all_results['micro_f1']))
all_trail_results['macro_f1'].append(np.mean(fold_all_results['macro_f1']))
test_result = {'micro_f1': np.mean(all_trail_results['micro_f1']),
'macro_f1': np.mean(all_trail_results['macro_f1'])}
test_max_result = {'micro_f1': np.max(all_trail_results['micro_f1']),
'macro_f1': np.max(all_trail_results['macro_f1'])}
test_min_result = {'micro_f1': np.min(all_trail_results['micro_f1']),
'macro_f1': np.min(all_trail_results['macro_f1'])}
all_results[noise_rate] = test_result
print(f"noise_rate {noise_rate}: {test_result}, max: {test_max_result}, min: {test_min_result}")
logging.info("Aug_1: {}, Aug_2: {}, noise_rate: {}: avg {}, max {}, min {}".format(augs_type[0], augs_type[1], noise_rate, test_result, test_max_result, test_min_result))
# results
# print(f'(E): Avg all folds test F1Mi={avg_results["micro_f1"]:.4f}, F1Ma={avg_results["macro_f1"]:.4f}')
logging.info("Aug_1: {}, Aug_2: {}, all_results: {}".format(augs_type[0], augs_type[1], all_results))
print("-"*90)
if __name__ == '__main__':
main()