-
Notifications
You must be signed in to change notification settings - Fork 2
/
util.py
353 lines (313 loc) · 14.6 KB
/
util.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
import os
import sys
import torch
import torchvision
import flwr as fl
import numpy as np
from torch.utils.data import Dataset
from tdc.utils import retrieve_label_name_list
from tdc.single_pred import Tox
from rdkit import Chem
import sklearn.metrics as metrics
def smiles_encoder(smiles, maxlen=386):
SMILES_CHARS = [' ',
'#', '%', '(', ')', '+', '-', '.', '/',
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9',
'=', '@',
'A', 'B', 'C', 'D', 'F', 'G', 'H', 'I', 'K', 'L', 'M', 'N', 'O', 'P',
'R', 'S', 'T', 'V', 'X', 'Y', 'Z',
'[', '\\', ']',
'a', 'b', 'c', 'd', 'e', 'g', 'i', 'l', 'n', 'o', 'p', 'r', 's',
't', 'u', 'y']
smi2index = dict((c, i) for i,c in enumerate(SMILES_CHARS))
smiles = Chem.MolToSmiles(Chem.MolFromSmiles(smiles))
one_hot = np.zeros((maxlen - 2, len(SMILES_CHARS)))
for i, c in enumerate(smiles):
one_hot[i, smi2index[c]] = 1
one_hot = np.concatenate([
np.zeros((1, len(SMILES_CHARS))),
one_hot,
np.zeros((1, len(SMILES_CHARS)))
])
return one_hot
def npz_loader(path):
sample = np.load(path)
x, y = torch.from_numpy(sample['x']), torch.from_numpy(sample['y'])
return (x, y)
def train(net, trainloader, epochs, device, flag):
"""Train the network."""
# Define loss and optimizer
if flag:
criterion = torch.nn.BCEWithLogitsLoss(reduction='none')
else:
criterion = torch.nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(net.parameters(), lr=0.001, momentum=0.9)
scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=1, gamma=0.7)
# Train the network
net.to(device)
net.train()
for epoch in range(epochs): # loop over the dataset multiple times
running_loss = 0.0
num_examples_train = 0
for data in trainloader:
if flag:
inputs, labels = data[0].float().to(device), data[1].float().to(device)
else:
inputs, labels = data[0].float().to(device), data[1].long().to(device)
num_examples_train += len(inputs)
# zero the parameter gradients
optimizer.zero_grad()
# forward + backward + optimize
outputs = net(inputs)
loss = criterion(outputs, labels)
if flag:
weight = torch.tensor([0.1, 0.9]).to(device)
weight_ = weight[labels.data.view(-1).long()].view_as(labels)
loss_class_weighted = loss * weight_
loss = loss_class_weighted.mean()
loss.backward()
optimizer.step()
# print statistics
running_loss += loss.item()
scheduler.step()
return num_examples_train, running_loss
def test(net, testloader, device, flag):
"""Validate the network on the entire test set."""
# Define loss and metrics
if flag:
criterion = torch.nn.BCEWithLogitsLoss()
else:
criterion = torch.nn.CrossEntropyLoss()
correct, total = 0, 0
loss = 0.0
num_examples_test = 0
# Evaluate the network
net.to(device)
net.eval()
with torch.no_grad():
if flag:
y_true, y_pred = [], []
for data in testloader:
if flag:
inputs, labels = data[0].float().to(device), data[1].float().to(device)
else:
inputs, labels = data[0].float().to(device), data[1].long().to(device)
num_examples_test += len(inputs)
outputs = net(inputs)
loss += criterion(outputs, labels).item()
total += labels.size(0)
if flag:
y_true.append(labels.detach().cpu())
y_pred.append(torch.nn.functional.softmax(outputs.detach().cpu()))
else:
_, predicted = torch.max(outputs.data, 1)
correct += (predicted == labels).sum().item()
accuracy = correct / total
else:
if flag:
y_true, y_pred = torch.cat(y_true, 0).numpy(), torch.cat(y_pred, 0).numpy()
fpr, tpr, threshold = metrics.roc_curve(y_true, y_pred)
metric = metrics.auc(fpr, tpr)
return loss, num_examples_test, metric
# adapted from my code: https://github.com/vaseline555/Federated-Averaging-PyTorch/blob/main/src/utils.py
class CustomTensorDataset(torch.utils.data.Dataset):
"""TensorDataset with support of transforms."""
def __init__(self, tensors, transform=None):
assert all(tensors[0].size(0) == tensor.size(0) for tensor in tensors)
self.tensors = tensors
self.transform = transform
def __getitem__(self, index):
x = self.tensors[0][index]
y = self.tensors[1][index]
if self.transform:
x = self.transform(x.numpy().astype(np.uint8))
return x, y
def __len__(self):
return self.tensors[0].size(0)
class CustomNumpyDataset(torch.utils.data.Dataset):
"""NumpyDataset with support of transforms."""
def __init__(self, path, train):
self.path = path
if train:
self.tensors = (
torch.from_numpy(np.load(os.path.join(path, "X_train.npy"))),
torch.from_numpy(np.load(os.path.join(path, "y_train.npy")))
)
else:
self.tensors = (
torch.from_numpy(np.load(os.path.join(path, "X_test.npy"))),
torch.from_numpy(np.load(os.path.join(path, "y_test.npy")))
)
self.data = self.tensors[0].squeeze().float()
self.targets = self.tensors[-1].float()
def __getitem__(self, index):
x = self.tensors[0][index]
y = self.tensors[1][index]
return x, y
def __len__(self):
return self.tensors[0].size(0)
def create_datasets(data_path, dataset_name, num_clients, num_shards, iid):
"""Split the whole dataset in IID or non-IID manner for distributing to clients."""
dataset_name = dataset_name.upper()
# get dataset from torchvision.datasets if exists
if hasattr(torchvision.datasets, dataset_name):
# set transformation differently per dataset
if dataset_name in ["CIFAR10"]:
transform = torchvision.transforms.Compose(
[
torchvision.transforms.ToTensor(),
torchvision.transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
]
)
elif dataset_name in ["MNIST"]:
transform = torchvision.transforms.ToTensor()
else:
# dataset not found exception
error_message = f"...dataset \"{dataset_name}\" cannot be found in TorchVision Datasets!"
raise AttributeError(error_message)
# prepare raw training & test datasets
training_dataset = torchvision.datasets.__dict__[dataset_name](
root=data_path,
train=True,
download=True,
transform=transform
)
test_dataset = torchvision.datasets.__dict__[dataset_name](
root=data_path,
train=False,
download=True,
transform=transform
)
# unsqueeze channel dimension for grayscale image datasets
if training_dataset.data.ndim == 3: # convert to NxHxW -> NxHxWx1
training_dataset.data.unsqueeze_(3)
num_categories = np.unique(training_dataset.targets).shape[0]
if "ndarray" not in str(type(training_dataset.data)):
training_dataset.data = np.asarray(training_dataset.data)
if "list" not in str(type(training_dataset.targets)):
training_dataset.targets = training_dataset.targets.tolist()
# split dataset according to iid flag
if iid:
# shuffle data
shuffled_indices = torch.randperm(len(training_dataset))
training_inputs = training_dataset.data[shuffled_indices]
training_labels = torch.Tensor(training_dataset.targets)[shuffled_indices]
# partition data into num_clients
split_size = len(training_dataset) // num_clients
split_datasets = list(
zip(
torch.split(torch.Tensor(training_inputs), split_size),
torch.split(torch.Tensor(training_labels), split_size)
)
)
# finalize bunches of local datasets
local_datasets = [
CustomTensorDataset(local_dataset, transform=transform)
for local_dataset in split_datasets
]
else:
# sort data by labels
sorted_indices = torch.argsort(torch.Tensor(training_dataset.targets))
training_inputs = training_dataset.data[sorted_indices]
training_labels = torch.Tensor(training_dataset.targets)[sorted_indices]
# partition data into shards first
shard_size = len(training_dataset) // num_shards #300
shard_inputs = list(torch.split(torch.Tensor(training_inputs), shard_size))
shard_labels = list(torch.split(torch.Tensor(training_labels), shard_size))
# sort the list to conveniently assign samples to each clients from at least two classes
shard_inputs_sorted, shard_labels_sorted = [], []
for i in range(num_shards // num_categories):
for j in range(0, ((num_shards // num_categories) * num_categories), (num_shards // num_categories)):
shard_inputs_sorted.append(shard_inputs[i + j])
shard_labels_sorted.append(shard_labels[i + j])
# finalize local datasets by assigning shards to each client
shards_per_clients = num_shards // num_clients
local_datasets = [
CustomTensorDataset(
(
torch.cat(shard_inputs_sorted[i:i + shards_per_clients]),
torch.cat(shard_labels_sorted[i:i + shards_per_clients]).long()
),
transform=transform
)
for i in range(0, len(shard_inputs_sorted), shards_per_clients)
]
# get custom dataset outside torchvision.datasets
else:
if dataset_name in ["TOX21"]:
if not os.path.exists(os.path.join(data_path, 'tox21')):
os.makedirs(os.path.join(data_path, 'tox21'))
label_list = retrieve_label_name_list('Tox21')
data = Tox(name='Tox21', path=data_path, label_name=label_list[0])
X_train, y_train = [], []
for idx, sample in data.get_split()['train'].loc[:, ['Drug', 'Y']].iterrows():
X_train.append(smiles_encoder(sample.Drug))
y_train.append(sample.Y)
for idx, sample in data.get_split()['valid'].loc[:, ['Drug', 'Y']].iterrows():
X_train.append(smiles_encoder(sample.Drug))
y_train.append(sample.Y)
np.save(os.path.join(data_path, f'tox21/X_train'), np.array(X_train))
np.save(os.path.join(data_path, f'tox21/y_train'), np.array(y_train))
X_test, y_test = [], []
for idx, sample in data.get_split()['test'].loc[:, ['Drug', 'Y']].iterrows():
X_test.append(smiles_encoder(sample.Drug))
y_test.append(sample.Y)
np.save(os.path.join(data_path, f'tox21/X_test'), np.array(X_test))
np.save(os.path.join(data_path, f'tox21/y_test'), np.array(y_test))
else:
print("Files already downloaded and verified")
else:
# dataset not found exception
error_message = f"...dataset \"{dataset_name}\" is not supported!"
raise AttributeError(error_message)
# prepare raw training & test datasets
training_dataset = CustomNumpyDataset(os.path.join(data_path, 'tox21'), train=True)
test_dataset = CustomNumpyDataset(os.path.join(data_path, 'tox21'), train=False)
# number of classes
num_categories = np.unique(training_dataset.targets).shape[0]
# split dataset according to iid flag
if iid:
# shuffle data
shuffled_indices = torch.randperm(len(training_dataset))
training_inputs = training_dataset.data[shuffled_indices]
training_labels = torch.Tensor(training_dataset.targets)[shuffled_indices]
# partition data into num_clients
split_size = len(training_dataset) // num_clients
split_datasets = list(
zip(
torch.split(torch.Tensor(training_inputs), split_size),
torch.split(torch.Tensor(training_labels), split_size)
)
)
# finalize bunches of local datasets
local_datasets = [
CustomTensorDataset(local_dataset)
for local_dataset in split_datasets
]
else:
# sort data by labels
sorted_indices = torch.argsort(torch.Tensor(training_dataset.targets))
training_inputs = training_dataset.data[sorted_indices]
training_labels = torch.Tensor(training_dataset.targets)[sorted_indices]
# partition data into shards first
shard_size = len(training_dataset) // num_shards #300
shard_inputs = list(torch.split(torch.Tensor(training_inputs), shard_size))
shard_labels = list(torch.split(torch.Tensor(training_labels), shard_size))
# sort the list to conveniently assign samples to each clients from at least two classes
shard_inputs_sorted, shard_labels_sorted = [], []
for i in range(num_shards // num_categories):
for j in range(0, ((num_shards // num_categories) * num_categories), (num_shards // num_categories)):
shard_inputs_sorted.append(shard_inputs[i + j])
shard_labels_sorted.append(shard_labels[i + j])
# finalize local datasets by assigning shards to each client
shards_per_clients = num_shards // num_clients
local_datasets = [
CustomTensorDataset(
(
torch.cat(shard_inputs_sorted[i:i + shards_per_clients]),
torch.cat(shard_labels_sorted[i:i + shards_per_clients])
),
)
for i in range(0, len(shard_inputs_sorted), shards_per_clients)
]
return local_datasets, test_dataset