def train_model(self):
self._model()
early_stopping = EarlyStopping(
monitor="val_loss",
patience=self.model_params["early_stopping_patience"],
verbose=1)
self.history = self.model.fit(
x=self.train_feature,
y=self.train_label,
epochs=self.model_params["epochs"],
verbose=self.model_params["verbose"],
class_weight=self.class_weights,
validation_data=(self.validation_feature, self.validation_label),
batch_size=self.model_params["batch_size"],
shuffle=self.model_params["shuffle"],
callbacks=[early_stopping])
# logger.info("History : ", self.history)
history_logger = self.history.history
for i in range(len(history_logger["acc"])):
logger.info(
"Epoch : {:<4d} loss : {:.5f} acc : {:.5f} val_loss : {:.5f} val_acc : {:.5f}"
.format(i, history_logger["loss"][i], history_logger["acc"][i],
history_logger["val_loss"][i],
history_logger["val_acc"][i]))
self._save_model()
self._save_figure(self.show_eval_figure)
self._model_evaluate()
self._get_predict_result_and_middle_feature()
generate_metrics(self.dic_path)
plot(self.dic_path, self.train_motor, self.test_motor)
def train_model(self):
self._model()
early_stopping = EarlyStopping(
monitor="val_loss",
patience=self.args.early_stopping_patience,
verbose=1)
self.history = self.model.fit(
x=[
self.data_dic['train_feature'],
self.data_dic['test_feature_for_transfer'],
self.data_dic['train_label'],
self.data_dic['test_label_for_transfer']
],
y=[
self.data_dic['train_label'],
self.data_dic['test_label_for_transfer']
],
epochs=self.args.epochs,
verbose=1,
class_weight=[self.class_weights, self.class_weights],
validation_data=([
self.data_dic['validation_feature'],
self.data_dic['validation_feature_for_transfer'],
self.data_dic['validation_label'],
self.data_dic['validation_label_for_transfer']
], [
self.data_dic['validation_label'],
self.data_dic['validation_label_for_transfer']
]),
batch_size=self.args.batch_size,
shuffle=True,
callbacks=[early_stopping])
self._save_model()
self._load_exist_model()
# self._save_figure(self.show_eval_figure)
# self._model_evaluate()
self._get_predict_result_and_middle_feature()
generate_metrics(self.dic_path)
plot(self.dic_path, self.args.train_motor, self.args.test_motor)
def train_model(self):
self._model()
early_stopping = EarlyStopping(
monitor="val_loss",
patience=self.args.early_stopping_patience,
verbose=1)
# data generator
training_generator = DataGenerator(
self.data_dic['train_feature'],
self.data_dic['test_feature_for_transfer'],
self.data_dic['train_label'],
self.data_dic['test_label_for_transfer'], self.model,
self.one_hot_encoder, self.args.batch_size)
validation_generator = DataGenerator(
self.data_dic['validation_feature'],
self.data_dic['validation_feature_for_transfer'],
self.data_dic['validation_label'],
self.data_dic['validation_label_for_transfer'], None, None,
self.args.batch_size)
self.history = self.model.fit_generator(
generator=training_generator,
validation_data=validation_generator,
use_multiprocessing=True,
workers=6,
epochs=self.args.epochs,
verbose=0,
class_weight=[self.class_weights, self.class_weights],
callbacks=[early_stopping])
self._save_model()
self._load_exist_model()
# self._save_figure(self.show_eval_figure)
# self._model_evaluate()
self._get_predict_result_and_middle_feature()
generate_metrics(self.dic_path)
plot(self.dic_path, self.args.train_motor, self.args.test_motor)
def predict_with_exist_model(self):
self._load_exist_model()
self._get_predict_result_and_middle_feature()
generate_metrics(self.dic_path)
plot(self.dic_path, self.train_motor, self.test_motor)
