-
Notifications
You must be signed in to change notification settings - Fork 1
/
func.py
435 lines (319 loc) · 15.6 KB
/
func.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
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
import plotly.graph_objects as go
import plotly.express as px
import matplotlib.pyplot as plt
import numpy as np
import torch
import itertools
import os
import random
import seaborn as sns
import torch
import mlflow
from matplotlib import pyplot as plt
import numpy as np
import torch.nn as nn
import mlflow.pytorch
import pandas as pd
from sklearn.metrics import f1_score
from sklearn.metrics import confusion_matrix, classification_report
import pyedflib
import scipy
def rename_columns(df):
# Приводит к правильному виду данные в df:
new_columns = []
for column in df.columns:
new_columns.append(column[:-4])
df.columns = new_columns
return df
def discrete_signal_resample(signal, time, new_sampling_rate):
## Производит ресемплирование
# Текущая частота дискретизации
current_sampling_rate = 1 / np.mean(np.diff(time))
# Количество точек в новой дискретизации
num_points_new = int(len(signal) * new_sampling_rate / current_sampling_rate)
# Используем scipy.signal.resample для изменения дискретизации
new_signal = scipy.signal.resample(signal, num_points_new)
new_time = np.linspace(time[0], time[-1], num_points_new)
return new_signal, new_time
def plot_signals_with_subplot(signal):
subplot_shape = (signal.shape[0],)
# subplot_shape - кортеж, указывающий количество строк в сетке subplot
# Проверяем, достаточно ли сигналов для создания всех подграфиков
if signal.shape[0] < subplot_shape[0]:
raise ValueError("Недостаточно сигналов для создания всех подграфиков в сетке subplot.")
# Создаем сетку subplot
fig, axes = plt.subplots(subplot_shape[0], figsize=(8, 7))
# Регулируем расположение графиков
plt.tight_layout()
# Строим графики для каждого сигнала
for i in range(subplot_shape[0]):
axes[i].plot(signal[i])
#axes[i].set_title(f'Signal {i+1}')
# Показываем графики
plt.show()
def get_full_signal(path, Fs_new=500, f_sreza=0.7):
# Открываем EDF файл
f = pyedflib.EdfReader(path)
# Получаем информацию о каналах
num_channels = f.signals_in_file
channels = f.getSignalLabels()
# Читаем данные по каналам
raw_data = []
for i in range(num_channels):
channel_data = f.readSignal(i)
raw_data.append(channel_data)
# Получаем частоту дискретизации
fd = f.getSampleFrequency(0)
# Закрываем файл EDF после чтения
f.close()
raw_data = np.array(raw_data)
# Создаем DataFrame
df = pd.DataFrame(data=raw_data.T,
index=range(raw_data.shape[1]),
columns=channels)
# Переименование столбцов при необходимости:
if 'ECG I-Ref' in df.columns:
df = rename_columns(df)
channels = df.columns
# Создание массива времени
Ts = 1/fd
t = []
for i in range(raw_data.shape[1]):
t.append(i*Ts)
# Ресемлинг:
df_new = pd.DataFrame()
for graph in channels:
sig = np.array(df[graph])
new_ecg, time_new = discrete_signal_resample(sig, t, Fs_new)
df_new[graph] = pd.Series(new_ecg)
df = df_new.copy()
# ФВЧ фильтрация артефактов дыхания:
df_new = pd.DataFrame()
for graph in channels:
sig = np.array(df[graph])
sos = scipy.signal.butter(1, f_sreza, 'hp', fs=Fs_new, output='sos')
avg = np.mean(sig)
filtered = scipy.signal.sosfilt(sos, sig)
filtered += avg
df_new[graph] = pd.Series(filtered)
df = df_new.copy()
# Выбор нужных столбцов
selected_columns = ['ECG I', 'ECG II', 'ECG V1', 'ECG V2', 'ECG V3', 'ECG V4', 'ECG V5', 'ECG V6']
selected_data = df[selected_columns]
# Преобразование данных в массив NumPy
numpy_array = selected_data.values.T
return numpy_array
def plot_confusion_matrix(cm, classes, normalize=False, title='Confusion matrix', cmap=plt.cm.Blues):
if normalize:
cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
plt.imshow(cm, interpolation='nearest', cmap=cmap)
plt.title(title)
plt.colorbar()
tick_marks = np.arange(len(classes))
plt.xticks(tick_marks, classes, rotation=45)
plt.yticks(tick_marks, classes)
fmt = '.2f' if normalize else 'd'
thresh = cm.max() / 2.
for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
plt.text(j, i, format(cm[i, j], fmt), horizontalalignment="center", color="white" if cm[i, j] > thresh else "black")
plt.tight_layout()
plt.ylabel('True label')
plt.xlabel('Predicted label')
def weighted_avg_f1(output, labels):
"""Функция расчета weighted avg F1-меры"""
# Преобразование списков в массивы numpy
output = np.array(output)
labels = np.array(labels)
# Создание тензоров PyTorch
output = torch.tensor(output)
labels = torch.tensor(labels)
predictions = torch.argmax(output, dim=1).cpu().numpy()
labels = labels.cpu().numpy()
weighted_f1 = f1_score(labels, predictions, average='weighted')
weighted_f1 = np.nan_to_num(weighted_f1, nan=0.0) # Замена NaN на 0 при делении на 0
return weighted_f1
def accuracy(output,labels):
"""Функция расчета accuracy"""
# Преобразование списков в массивы numpy
output = np.array(output)
labels = np.array(labels)
# Создание тензоров PyTorch
output = torch.tensor(output)
labels = torch.tensor(labels)
predictions = torch.argmax(output,dim=1)
correct = (predictions == labels).sum().cpu().numpy()
return correct / len(labels)
def evaluate_model(model, data_loader):
"""Функция для логирования артефактов в MLflow"""
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)
model.eval()
true_labels = []
predicted_labels = []
with torch.no_grad():
for data in data_loader:
inputs, labels = data['signal'].to(device).float(), data['category'].to(device)
outputs = model(inputs)
_, predicted = torch.max(outputs, 1)
true_labels.extend(labels.cpu().numpy())
predicted_labels.extend(predicted.cpu().numpy())
# Compute confusion matrix
cm = confusion_matrix(true_labels, predicted_labels)
# Сохранение отчета в текстовый файл
report = classification_report(true_labels, predicted_labels)
output_file = "classification_report.txt"
with open(output_file, "w") as f:
f.write(report)
mlflow.log_artifact("classification_report.txt")
os.remove("classification_report.txt")
# Save confusion matrix as CSV artifact
df = pd.DataFrame(cm)
new_columns = [f"predicted class {i}" for i in range(df.shape[1])]
# Установка новых названий столбцов
df.columns = new_columns
# Добавление индексов
df.insert(0, "real\pred", [f"real class {i}" for i in range(df.shape[0])])
# Сохранение измененной таблицы
df.to_csv("confusion_matrix.csv", index=False)
mlflow.log_artifact("confusion_matrix.csv")
os.remove("confusion_matrix.csv")
def train_model(model, dataloader_train, dataloader_val, batch_size,
name_save, start_weight=None,
name_experiment=None, lr=1e-4, epochs=100,
scheduler=True, scheduler_step_size=10, dataset_name=None,
f_sampling=500, seed=42, n_points=2048, num_channels=8,
filt='ФВЧ 0.7 Гц', gamma=0.5, noise_std=0):
"""Обучение классификационной сверточной сети
Args:
model: Класс модели pytorch
dataloader_train: Обучающий даталоудер
dataloader_val: Валидационный даталоудер
batch_size: Размер одного батча
name_save: Имя модели для сохранения в папку models
start_weight: Если указать веса, то сеть будет в режиме fine tune. Defaults to None.
name_experiment: Имя эксперимента для MLflow. Нужно при mlflow_tracking=True. Defaults to None.
lr: Скорость обучения. Defaults to 1e-4.
epochs: Число эпох обучения. Defaults to 100.
scheduler (bool): Включение/выключение lr шедулера. Defaults to True.
scheduler_step_size (int): Шаг шедулера при scheduler=True. Defaults to 10.
dataset_name: Имя датасета для логирования в MLflow. Defaults to None.
seed (int): Seed рандома. Defaults to 42.
filt: Вид фильтрации
gamma: Величина коэффициента lr шедулера
noise_std: Величина std шума на трейне
"""
device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
print(f'Обучение будет производиться на {device}')
torch.manual_seed(seed)
torch.cuda.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
np.random.seed(seed)
random.seed(seed)
torch.backends.cudnn.benchmark = False
torch.backends.cudnn.deterministic = True
if dataset_name != None:
dataset_name = dataset_name.split('/')[-1]
directory_save = 'models'
if not os.path.exists(directory_save):
os.makedirs(directory_save)
if name_experiment == None:
name_experiment = name_save
with mlflow.start_run(run_name=name_experiment) as run:
mlflow.log_param("Model", model.__class__.__name__)
if start_weight != None:
model.load_state_dict(torch.load(start_weight))
optimizer = torch.optim.Adam(model.parameters(), lr=lr)
if scheduler:
gamma_val = gamma
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer=optimizer,
step_size=scheduler_step_size,
gamma=gamma_val)
model = model.to(device)
mlflow.log_param("filt", filt)
mlflow.log_param("Training random noise std", noise_std)
mlflow.log_param("Input shape", f'torch.Size([batch_size, {num_channels}, {n_points}])')
mlflow.log_param("F sampling ECG", f_sampling)
mlflow.log_param("Points samping", n_points)
if scheduler:
mlflow.log_param("scheduler", 'On')
mlflow.log_param("scheduler_step_size", scheduler_step_size)
mlflow.log_param("scheduler_gamma", gamma_val)
else:
mlflow.log_param("scheduler", 'Off')
mlflow.log_param("lr", lr)
mlflow.log_param("optimizer", 'Adam')
mlflow.log_param("epochs", epochs)
mlflow.log_param("loss", 'CrossEntropyLoss')
mlflow.log_param("batch_size", batch_size)
mlflow.log_param("dataset", dataset_name)
mlflow.log_param("seed", seed)
if start_weight != None:
mlflow.log_param("Fine-tuning", True)
else:
mlflow.log_param("Fine-tuning", False)
max_epoch_f1_val = 0
loss_func = nn.CrossEntropyLoss()
for epoch in range(epochs):
model.train()
running_loss = 0.0
all_outputs = []
all_targets = []
for i, data in enumerate(dataloader_train, 0):
inputs, labels = data['signal'].to(device).float(), data['category'].to(device)
optimizer.zero_grad()
outputs = model(inputs)
loss = loss_func(outputs, labels)
loss.backward()
optimizer.step()
running_loss += loss.item()
all_outputs.extend(outputs.cpu().detach().numpy())
all_targets.extend(labels.cpu().numpy())
train_epoch_loss = running_loss / len(dataloader_train)
train_epoch_acc = accuracy(all_outputs, all_targets)
train_epoch_f1 = weighted_avg_f1(all_outputs, all_targets)
mlflow.log_metric("train_epoch_accuracy", train_epoch_acc, step=(epoch+1))
mlflow.log_metric("train_epoch_loss", train_epoch_loss, step=(epoch+1))
mlflow.log_metric("train_epoch_f1", train_epoch_f1, step=(epoch+1))
# validation
model.eval()
with torch.no_grad():
running_loss = 0.0
all_outputs = []
all_targets = []
for data in dataloader_val:
inputs, labels = data['signal'].to(device).float(), data['category'].to(device)
outputs = model(inputs)
loss = loss_func(outputs, labels)
running_loss += loss.item()
all_outputs.extend(outputs.cpu().detach().numpy())
all_targets.extend(labels.cpu().numpy())
val_epoch_loss = running_loss / len(dataloader_val)
val_epoch_acc = accuracy(all_outputs, all_targets)
val_epoch_f1 = weighted_avg_f1(all_outputs, all_targets)
mlflow.log_metric("validation_epoch_accuracy", val_epoch_acc, step=(epoch+1))
mlflow.log_metric("validation_epoch_loss", val_epoch_loss, step=(epoch+1))
mlflow.log_metric("validation_epoch_f1", val_epoch_f1, step=(epoch+1))
if scheduler:
lr_scheduler.step()
# Вывод значения функции потерь на каждой 5 эпохе
if ((epoch+1) % 5 == 0) or epoch==0:
print(f'Epoch {epoch+1}/{epochs}, Train Loss: {train_epoch_loss:.4f},'
f' Train Aсс: {train_epoch_acc:.4f}'
f' Val Loss: {val_epoch_loss:.4f}, Val Acc:{val_epoch_acc:.4f} ')
if epoch >= 1 and val_epoch_f1 > max_epoch_f1_val:
max_epoch_f1_val = val_epoch_f1
acc_model = val_epoch_acc
model_to_save = model
epoch_best = epoch + 1
name_save_model = directory_save + '/' + name_save +'.pth'
torch.save(model_to_save.state_dict(), name_save_model)
evaluate_model(model=model, data_loader=dataloader_val)
print('Обучение завершено')
print(f'Сохранена модель {name_save_model} с лучшим weighted avg f1 на валидации = {max_epoch_f1_val}')
print('Accuracy данной модели равно', acc_model)
mlflow.log_metric("max f1 saved model", max_epoch_f1_val)
mlflow.log_metric("accuracy of model", acc_model)
mlflow.log_metric("epoch of save", epoch_best)
mlflow.log_artifact(name_save_model)
#mlflow.log_artifact('model.py')
###########################################