def train_model(filename_train_validation_set,
filename_labels_train_validation_set, filename_sample_weights,
filter_density1, filter_density2, pool_n_row, pool_n_col,
dropout, input_shape, file_path_model, filename_log):
"""
train final model save to model path
"""
filenames_train, Y_train, sample_weights_train, \
filenames_validation, Y_validation, sample_weights_validation, \
filenames_features, Y_train_validation, sample_weights, class_weights = \
load_data_jingju(filename_labels_train_validation_set,
filename_sample_weights)
model_0 = jordi_model(filter_density1, filter_density2, pool_n_row,
pool_n_col, dropout, input_shape, 'temporal')
batch_size = 128
patience = 10
print(model_0.count_params())
model_train(model_0, batch_size, patience, input_shape,
filename_train_validation_set, filenames_train, Y_train,
sample_weights_train, filenames_validation, Y_validation,
sample_weights_validation, filenames_features,
Y_train_validation, sample_weights, class_weights,
file_path_model, filename_log)
def train_model(filename_train_validation_set,
filename_labels_train_validation_set, filename_sample_weights,
filter_density, dropout, input_shape, file_path_model,
filename_log):
"""
train final model save to model path
"""
filenames_train, Y_train, sample_weights_train, \
filenames_validation, Y_validation, sample_weights_validation, \
filenames_features, Y_train_validation, sample_weights, class_weights = \
load_data_jingju(filename_labels_train_validation_set,
filename_sample_weights)
model_0 = jan(filter_density=filter_density,
dropout=dropout,
input_shape=input_shape,
batchNorm=True)
batch_size = 128
patience = 10
print(model_0.count_params())
model_train(model_0, batch_size, patience, input_shape,
filename_train_validation_set, filenames_train, Y_train,
sample_weights_train, filenames_validation, Y_validation,
sample_weights_validation, filenames_features,
Y_train_validation, sample_weights, class_weights,
file_path_model, filename_log)
def finetune_model_validation(filename_train_validation_set,
filename_labels_train_validation_set,
filename_sample_weights,
filter_density,
dropout,
input_shape,
file_path_model,
filename_log,
model_name,
path_model,
channel=1):
"""
train model with validation
"""
filenames_train, Y_train, sample_weights_train, \
filenames_validation, Y_validation, sample_weights_validation, \
filenames_features, Y_train_validation, sample_weights, class_weights = \
load_data_jingju(filename_labels_train_validation_set,
filename_sample_weights)
# load pretrained model
model_pretrained = load_model(filepath=path_model)
if model_name == 'retrained':
model = model_pretrained
multi_inputs = False
else:
model = model_switcher(model_name=model_name,
filter_density=filter_density,
dropout=dropout,
input_shape=input_shape,
channel=channel,
model_pretrained=model_pretrained,
activation_dense='sigmoid')
multi_inputs = True
batch_size = 256
patience = 15
model_train_validation(model,
batch_size,
patience,
input_shape,
filename_train_validation_set,
filenames_train,
Y_train,
sample_weights_train,
filenames_validation,
Y_validation,
sample_weights_validation,
class_weights,
file_path_model,
filename_log,
channel,
multi_inputs=multi_inputs)
def train_model_validation(filename_train_validation_set,
filename_labels_train_validation_set,
filename_sample_weights,
filter_density,
dropout,
input_shape,
file_path_model,
filename_log,
model_name='baseline',
activation_dense='sigmoid',
channel=1):
"""
train model with validation
"""
filenames_train, Y_train, sample_weights_train, \
filenames_validation, Y_validation, sample_weights_validation, \
filenames_features, Y_train_validation, sample_weights, class_weights = \
load_data_jingju(filename_labels_train_validation_set,
filename_sample_weights)
model_0 = model_switcher(model_name=model_name,
filter_density=filter_density,
dropout=dropout,
input_shape=input_shape,
channel=channel,
activation_dense=activation_dense)
batch_size = 256
patience = 15
model_train_validation(model_0, batch_size, patience, input_shape,
filename_train_validation_set, filenames_train,
Y_train, sample_weights_train, filenames_validation,
Y_validation, sample_weights_validation,
class_weights, file_path_model, filename_log,
channel)