def train(args, params):
print("Loading training set...")
train = load.load_dataset(params['train'])
print("Loading dev set...")
dev = load.load_dataset(params['dev'])
print("Building preprocessor...")
preproc = load.Preproc(*train)
print("train_set_classes:", preproc.classes)
print("Training size: " + str(len(train[0])) + " examples.")
print("Dev size: " + str(len(dev[0])) + " examples.")
save_dir = make_save_dir(params['save_dir'], args.experiment)
util.save(preproc, save_dir)
params.update({
"input_shape": [None, 1],
"num_categories": len(preproc.classes)
})
model = network.build_network(**params)
stopping = keras.callbacks.EarlyStopping(patience=30)
reduce_lr = keras.callbacks.ReduceLROnPlateau(
factor=0.1, patience=2, min_lr=params["learning_rate"] * 0.001)
checkpointer = keras.callbacks.ModelCheckpoint(
filepath=get_filename_for_saving(save_dir), save_best_only=False)
batch_size = params.get("batch_size", 32)
# summary = str(model.summary(print_fn=lambda x: fh.write(x + '\n')))
# out = open("/content/ecg/report.txt",'w')
# out.write(summary)
# out.close
if params.get("generator", False):
train_gen = load.data_generator(batch_size, preproc, *train)
dev_gen = load.data_generator(batch_size, preproc, *dev)
model.fit_generator(train_gen,
steps_per_epoch=int(len(train[0]) / batch_size),
epochs=MAX_EPOCHS,
validation_data=dev_gen,
validation_steps=int(len(dev[0]) / batch_size),
callbacks=[checkpointer, reduce_lr, stopping])
# util.learning_curve(history)
else:
train_x, train_y = preproc.process(*train)
dev_x, dev_y = preproc.process(*dev)
model.fit(train_x,
train_y,
batch_size=batch_size,
epochs=MAX_EPOCHS,
validation_data=(dev_x, dev_y),
callbacks=[checkpointer, reduce_lr, stopping])
def train(args, params):
print("Loading training set...")
train = load.load_dataset(params['train'])
print("Loading dev set...")
dev = load.load_dataset(params['dev'])
print("Building preprocessor...")
preproc = load.Preproc(*train)
print("Training size: " + str(len(train[0])) + " examples.")
print("Dev size: " + str(len(dev[0])) + " examples.")
save_dir = make_save_dir(params['save_dir'], args.experiment)
util.save(preproc, save_dir)
params.update({
"input_shape": [None, 1],
"num_categories": len(preproc.classes)
})
print(params)
model = network.build_network(**params)
stopping = keras.callbacks.EarlyStopping(patience=8)
reduce_lr = keras.callbacks.ReduceLROnPlateau(
factor=0.1, patience=2, min_lr=params["learning_rate"] * 0.001)
checkpointer = keras.callbacks.ModelCheckpoint(
filepath=get_filename_for_saving(save_dir), save_best_only=False)
ckpt_best = keras.callbacks.ModelCheckpoint(os.path.join(
save_dir, 'best.hdf5'),
save_best_only=True)
batch_size = params.get("batch_size", 32)
if params.get("generator", False):
train_gen = load.data_generator(batch_size, preproc, *train)
dev_gen = load.data_generator(batch_size, preproc, *dev)
model.fit_generator(
train_gen,
steps_per_epoch=int(len(train[0]) / batch_size),
epochs=MAX_EPOCHS,
validation_data=dev_gen,
validation_steps=int(len(dev[0]) / batch_size),
callbacks=[checkpointer, ckpt_best, reduce_lr, stopping])
else:
train_x, train_y = preproc.process(*train)
dev_x, dev_y = preproc.process(*dev)
model.fit(train_x,
train_y,
batch_size=batch_size,
epochs=MAX_EPOCHS,
validation_data=(dev_x, dev_y),
callbacks=[checkpointer, ckpt_best, reduce_lr, stopping])
def predict(data_file, model_path):
ecg = load.load_ecg(data_file)
preproc = load.Preproc(ecg, [])
x = preproc.process_x([ecg])
#print(x.shape)
model = keras.models.load_model(model_path)
probs = model.predict(x, verbose=1)
#print(probs.shape)
return probs[0]
from keras.models import Model
import scipy.io as scio
from sklearn.metrics import confusion_matrix, accuracy_score, recall_score, precision_score, f1_score
MAX_EPOCHS = 160
batch_size = 32
if __name__ == '__main__':
params = util.config()
save_dir = params['save_dir']
print("Loading training set...")
train = load.load_dataset(params['train'])
print("Loading dev set...")
dev = load.load_dataset(params['dev'])
print("Building preprocessor...")
preproc = load.Preproc(*train)
print("Training size: " + str(len(train[0])) + " examples.")
print("Dev size: " + str(len(dev[0])) + " examples.")
params.update({
"input_shape": [8000, 1],
"num_categories": len(preproc.classes)
})
#create the cl-pcg-net
model = network.build_network(**params)
#learning rate reduce strategy
def scheduler(epoch):
if epoch % 80 == 0 and epoch != 0:
lr = K.get_value(model.optimizer.lr)
def train(args, params):
print("Loading training set...")
train = load.load_dataset(params['train'])
print("Loading dev set...")
dev = load.load_dataset(params['dev'])
print("Building preprocessor...")
preproc = load.Preproc(*train)
print("Training size: " + str(len(train[0])) + " examples.")
print("Dev size: " + str(len(dev[0])) + " examples.")
preproc.set_window_size(params['window_size'])
save_dir = make_save_dir(params['save_dir'], args.experiment)
util.save(preproc, save_dir)
print('classes', preproc.classes)
print("Skip Init initial Alpha : {}, L2 : {}, Lr : {}, batch_size : {}".
format(params["skip_init_a"], params["conv_l2"],
params["learning_rate"], params["batch_size"]))
params.update({
"input_shape": [None, 1],
"num_categories": len(preproc.classes)
})
model = network.build_network(**params)
print(model.summary())
if params.get('is_regular_conv', False):
plot_model(model, to_file='model_regular_conv.png', show_shapes=True)
else:
if params.get('conv_batch_norm'):
plot_model(model,
to_file='model_residual_conv_bn.png',
show_shapes=True)
else:
plot_model(model,
to_file='model_residual_conv_nobn.png',
show_shapes=True)
stopping = keras.callbacks.EarlyStopping(patience=8)
reduce_lr = keras.callbacks.ReduceLROnPlateau(
factor=0.1, patience=2, min_lr=params["learning_rate"] * 0.001)
# checkpointer = keras.callbacks.ModelCheckpoint(
# filepath=get_filename_for_saving(save_dir),
# save_best_only=False, save_weights_only=False)
checkpointer = keras.callbacks.ModelCheckpoint(
filepath=get_filename_for_saving(save_dir),
save_weights_only=False,
save_best_only=True,
monitor='val_accuracy',
mode='max',
)
batch_size = params.get("batch_size", 32)
if params.get("generator", False):
train_gen = load.data_generator(batch_size, preproc, *train)
dev_gen = load.data_generator(batch_size, preproc, *dev)
train_gen = np.array(train_gen)
dev_gen = np.array(dev_gen)
# print('train_gen m, s: ', train_gen.mean(), train_gen.std())
# print('train_gen min, max: ', train_gen.min(), train_gen.max())
print('train_gen shape : ', train_gen.shape)
print('dev_gen shape : ', train_gen.shape)
model.fit_generator(train_gen,
steps_per_epoch=int(len(train[0]) / batch_size),
epochs=params.get("epoch", 10),
validation_data=dev_gen,
validation_steps=int(len(dev[0]) / batch_size),
callbacks=[checkpointer, reduce_lr, stopping])
else:
train_x, train_y = preproc.process(*train)
dev_x, dev_y = preproc.process(*dev)
print('train_x original shape : ', train_x.shape)
print('train_y original shape : ', train_y.shape)
print('test_x original shape : ', dev_x.shape)
print('test_y original shape : ', dev_y.shape)
# train_x = np.array(train_x)
# train_y = np.array(train_y)
# test_x = np.array(dev_x)
# test_y = np.array(dev_y)
window_size = 256
n_sample = 40
r_i = 0
r_length = 1
# print('random number : ', np.random.choice(test_x.shape[1] // window_size, 1)[0])
# train_x = train_x[:, 256 * r_i:256 * (r_i+1)* r_length, :]
# train_y = train_y[:, r_i, :]
#
# test_x = test_x[:, 256 * r_i:256 * (r_i+1) * r_length, :]
# test_y = test_y[:, r_i, :]
# train_x = np.array(all_train_x)
# train_y = np.array(all_train_y)
# FOR PLAINTEXT conversion start
# get 8 windows
# train_x = train_x[:, :window_size * n_sample, :]
# train_y = train_y[:, r_i, :]
# test_x = test_x[:, :window_size * n_sample, :]
# test_y = test_y[:, r_i, :]
# test_x = np.array(random_test_x)
# train_x = np.array(train_x)
#
# train_x = np.squeeze(train_x, axis=(2,))
# test_x = np.squeeze(test_x, axis=(2,))
#
# print('train_x m, s: ', train_x.mean(), train_x.std())
# print('train_x min, max: ', train_x.min(), train_x.max())
#
# print('train_x shape : ', train_x.shape)
# print('train_y shape : ', train_y.shape)
# print('test_x shape : ', test_x.shape)
# print('test_y shape : ', test_y.shape)
# gt = np.argmax(test_y, axis=1)
# print('0:', len(gt[gt == 0]))
# print('1:', len(gt[gt == 1]))
# print('2:', len(gt[gt == 2]))
# print('3:', len(gt[gt == 3]))
# print('gt shape:', gt.shape)
# data_dir = 'processed_data_2560'
# data_dir = 'processed_data_5120'
#
# if not os.path.exists(data_dir):
# os.makedirs(data_dir)
# train_file_suffix = 'train'
# test_file_suffix = 'test'
#
# file_name_train_x = 'X{}'.format(train_file_suffix)
# file_name_train_y = 'y{}'.format(train_file_suffix)
# file_name_test_x = 'X{}'.format(test_file_suffix)
# file_name_test_y = 'y{}'.format(test_file_suffix)
#
# np.savetxt('{}/{}'.format(data_dir, file_name_train_x), train_x, delimiter=',', fmt='%1.8f')
# np.savetxt('{}/{}'.format(data_dir, file_name_train_y), train_y, delimiter=',', fmt='%1.8f')
# np.savetxt('{}/{}'.format(data_dir, file_name_test_x), test_x, delimiter=',', fmt='%1.8f')
# np.savetxt('{}/{}'.format(data_dir, file_name_test_y), test_y, delimiter=',', fmt='%1.8f')
# print('train_x shape : ', train_x.shape)
# print('train_y shape : ', train_y.shape)
# print('test_x shape : ', test_x.shape)
# print('test_y shape : ', test_y.shape)
model.fit(train_x,
train_y,
batch_size=batch_size,
epochs=params.get("epoch", 10),
validation_data=(dev_x, dev_y),
callbacks=[checkpointer, reduce_lr, stopping])
def train(args, params):
print("Loading training set...")
train = load.load_dataset(params['train'])
print("Loading dev set...")
dev = load.load_dataset(params['dev'])
print("Building preprocessor...")
preproc = load.Preproc(*train)
print("Training size: " + str(len(train[0])) + " examples.")
print("Dev size: " + str(len(dev[0])) + " examples.")
save_dir = make_save_dir(params['save_dir'], args.experiment)
util.save(preproc, save_dir)
params.update({
"input_shape": [None, 1],
"num_categories": len(preproc.classes)
})
'''
<class 'dict'>:
{'conv_subsample_lengths': [1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2],
'conv_filter_length': 16,
'conv_num_filters_start': 32,
'conv_init': 'he_normal',
'conv_activation': 'relu',
'conv_dropout': 0.2,
'conv_num_skip': 2,
'conv_increase_channels_at': 4,
'learning_rate': 0.001,
'batch_size': 32,
'train': '../examples/cinc17/train.json',
'dev': '../examples/cinc17/dev.json',
'generator': True,
'save_dir': 'saved',
'input_shape': [None, 1],
'num_categories': 4}
'''
model = network.build_network(**params)
stopping = keras.callbacks.EarlyStopping(patience=8)
reduce_lr = keras.callbacks.ReduceLROnPlateau(
factor=0.1, patience=2, min_lr=params["learning_rate"] * 0.001)
checkpointer = keras.callbacks.ModelCheckpoint(
filepath=get_filename_for_saving(save_dir), save_best_only=False)
batch_size = params.get("batch_size", 32)
#把网络结构打出来
# model.summary()
if params.get("generator", False):
train_gen = load.data_generator(batch_size, preproc, *train)
dev_gen = load.data_generator(batch_size, preproc, *dev)
model.fit_generator(train_gen,
steps_per_epoch=int(len(train[0]) / batch_size),
epochs=MAX_EPOCHS,
validation_data=dev_gen,
validation_steps=int(len(dev[0]) / batch_size),
callbacks=[checkpointer, reduce_lr, stopping])
else:
train_x, train_y = preproc.process(*train)
dev_x, dev_y = preproc.process(*dev)
model.fit(train_x,
train_y,
batch_size=batch_size,
epochs=MAX_EPOCHS,
validation_data=(dev_x, dev_y),
callbacks=[checkpointer, reduce_lr, stopping])
#保存训练结果
model.save('my_model.h5')
#返回训练之后的模型
return model
