# Supervised fine tuning parameters
flags.DEFINE_string('finetune_act_func', 'sigmoid', 'Activation function.')
flags.DEFINE_float('finetune_learning_rate', 0.9, 'Learning rate.')
flags.DEFINE_float('finetune_momentum', 0.7, 'Momentum parameter.')
flags.DEFINE_integer('finetune_num_epochs', 100, 'Number of epochs.')
flags.DEFINE_integer('finetune_batch_size', 100, 'Size of each mini-batch.')
flags.DEFINE_string('finetune_opt', 'gradient_descent',
'["gradient_descent", "ada_grad", "momentum", "adam"]')
flags.DEFINE_string(
'finetune_loss_func', 'softmax_cross_entropy',
'Loss function. ["mean_squared", "softmax_cross_entropy"]')
flags.DEFINE_float('finetune_dropout', 1, 'Dropout parameter.')
# Conversion of Autoencoder layers parameters from string to their specific type
rbm_layers = utilities.flag_to_list(FLAGS.rbm_layers, 'int')
rbm_learning_rate = utilities.flag_to_list(FLAGS.rbm_learning_rate, 'float')
rbm_num_epochs = utilities.flag_to_list(FLAGS.rbm_num_epochs, 'int')
rbm_batch_size = utilities.flag_to_list(FLAGS.rbm_batch_size, 'int')
rbm_gibbs_k = utilities.flag_to_list(FLAGS.rbm_gibbs_k, 'int')
# Parameters validation
assert FLAGS.dataset in ['mnist', 'cifar10', 'custom']
assert FLAGS.finetune_act_func in ['sigmoid', 'tanh', 'relu']
assert FLAGS.finetune_loss_func in ['mean_squared', 'softmax_cross_entropy']
assert len(rbm_layers) > 0
if __name__ == '__main__':
utilities.random_seed_np_tf(FLAGS.seed)
'finetune_enc_act_func', 'relu,',
'Activation function for the encoder fine-tuning phase. ["sigmoid, "tanh", "relu"]'
)
flags.DEFINE_string(
'finetune_dec_act_func', 'sigmoid,',
'Activation function for the decoder fine-tuning phase. ["sigmoid, "tanh", "relu"]'
)
flags.DEFINE_integer('finetune_num_epochs', 10, 'Number of epochs.')
flags.DEFINE_integer('finetune_batch_size', 10, 'Size of each mini-batch.')
flags.DEFINE_string('finetune_opt', 'sgd',
'["sgd", "ada_grad", "momentum", "adam"]')
flags.DEFINE_string('finetune_loss_func', 'mse', 'Loss function.')
flags.DEFINE_float('finetune_dropout', 1, 'Dropout parameter.')
# Conversion of Autoencoder layers parameters from string to their specific type
rbm_names = utilities.flag_to_list(FLAGS.rbm_names, 'str')
rbm_layers = utilities.flag_to_list(FLAGS.rbm_layers, 'int')
rbm_noise = utilities.flag_to_list(FLAGS.rbm_noise, 'str')
rbm_learning_rate = utilities.flag_to_list(FLAGS.rbm_learning_rate, 'float')
rbm_num_epochs = utilities.flag_to_list(FLAGS.rbm_num_epochs, 'int')
rbm_batch_size = utilities.flag_to_list(FLAGS.rbm_batch_size, 'int')
rbm_gibbs_k = utilities.flag_to_list(FLAGS.rbm_gibbs_k, 'int')
finetune_enc_act_func = utilities.flag_to_list(FLAGS.finetune_enc_act_func,
'str')
finetune_dec_act_func = utilities.flag_to_list(FLAGS.finetune_dec_act_func,
'str')
# Parameters validation
assert FLAGS.dataset in ['mnist', 'cifar10', 'custom']
assert len(rbm_layers) > 0
flags.DEFINE_string('rbm_num_epochs', '10,', 'Number of epochs.')
flags.DEFINE_string('rbm_batch_size', '32,', 'Size of each mini-batch.')
flags.DEFINE_string('rbm_gibbs_k', '1,', 'Gibbs sampling steps.')
# Supervised fine tuning parameters
flags.DEFINE_string('finetune_act_func', 'relu', 'Activation function.')
flags.DEFINE_float('finetune_learning_rate', 0.01, 'Learning rate.')
flags.DEFINE_float('finetune_momentum', 0.9, 'Momentum parameter.')
flags.DEFINE_integer('finetune_num_epochs', 10, 'Number of epochs.')
flags.DEFINE_integer('finetune_batch_size', 32, 'Size of each mini-batch.')
flags.DEFINE_string('finetune_opt', 'momentum', '["sgd", "ada_grad", "momentum", "adam"]')
flags.DEFINE_string('finetune_loss_func', 'softmax_cross_entropy', 'Loss function. ["mse", "softmax_cross_entropy"]')
flags.DEFINE_float('finetune_dropout', 1, 'Dropout parameter.')
# Conversion of Autoencoder layers parameters from string to their specific type
rbm_layers = utilities.flag_to_list(FLAGS.rbm_layers, 'int')
rbm_learning_rate = utilities.flag_to_list(FLAGS.rbm_learning_rate, 'float')
rbm_num_epochs = utilities.flag_to_list(FLAGS.rbm_num_epochs, 'int')
rbm_batch_size = utilities.flag_to_list(FLAGS.rbm_batch_size, 'int')
rbm_gibbs_k = utilities.flag_to_list(FLAGS.rbm_gibbs_k, 'int')
# Parameters validation
assert FLAGS.dataset in ['mnist', 'cifar10', 'custom']
assert FLAGS.finetune_act_func in ['sigmoid', 'tanh', 'relu']
assert len(rbm_layers) > 0
if __name__ == '__main__':
utilities.random_seed_np_tf(FLAGS.seed)
if FLAGS.dataset == 'mnist':
# Autoencoder layers specific parameters
flags.DEFINE_string('dae_layers', '256,', 'Comma-separated values for the layers in the sdae.')
flags.DEFINE_string('dae_l2reg', '5e-4,', 'Regularization parameter for the autoencoders. If 0, no regularization.')
flags.DEFINE_string('dae_enc_act_func', 'sigmoid,', 'Activation function for the encoder. ["sigmoid", "tanh"]')
flags.DEFINE_string('dae_dec_act_func', 'none,', 'Activation function for the decoder. ["sigmoid", "tanh", "none"]')
flags.DEFINE_string('dae_loss_func', 'mean_squared,', 'Loss function. ["mean_squared" or "cross_entropy"]')
flags.DEFINE_string('dae_opt', 'gradient_descent,', '["gradient_descent", "ada_grad", "momentum", "adam"]')
flags.DEFINE_string('dae_learning_rate', '0.01,', 'Initial learning rate.')
flags.DEFINE_string('dae_num_epochs', '10,', 'Number of epochs.')
flags.DEFINE_string('dae_batch_size', '10,', 'Size of each mini-batch.')
flags.DEFINE_string('dae_corr_type', 'none,', 'Type of input corruption. ["none", "masking", "salt_and_pepper"]')
flags.DEFINE_string('dae_corr_frac', '0.0,', 'Fraction of the input to corrupt.')
# Conversion of Autoencoder layers parameters from string to their specific type
dae_layers = utilities.flag_to_list(FLAGS.dae_layers, 'int')
dae_enc_act_func = utilities.flag_to_list(FLAGS.dae_enc_act_func, 'str')
dae_dec_act_func = utilities.flag_to_list(FLAGS.dae_dec_act_func, 'str')
dae_opt = utilities.flag_to_list(FLAGS.dae_opt, 'str')
dae_loss_func = utilities.flag_to_list(FLAGS.dae_loss_func, 'str')
dae_learning_rate = utilities.flag_to_list(FLAGS.dae_learning_rate, 'float')
dae_l2reg = utilities.flag_to_list(FLAGS.dae_l2reg, 'float')
dae_corr_type = utilities.flag_to_list(FLAGS.dae_corr_type, 'str')
dae_corr_frac = utilities.flag_to_list(FLAGS.dae_corr_frac, 'float')
dae_num_epochs = utilities.flag_to_list(FLAGS.dae_num_epochs, 'int')
dae_batch_size = utilities.flag_to_list(FLAGS.dae_batch_size, 'int')
# Parameters validation
assert all([0. <= cf <= 1. for cf in dae_corr_frac])
assert all([ct in ['masking', 'salt_and_pepper', 'none'] for ct in dae_corr_type])
assert FLAGS.dataset in ['mnist', 'cifar10', 'custom']
'rbm_layers', '100,',
'Comma-separated values for the number of hidden units in the layers.')
flags.DEFINE_boolean('finetuned', False,
'default False, Whether to load a finetuned model.')
flags.DEFINE_string('samples', None, 'Npy file with the test data.')
flags.DEFINE_boolean(
'memory_mapping', True,
'default True, Whether to use memory mapping to avoid loading the full dataset into RAM.'
)
flags.DEFINE_boolean(
'load_batch', False,
'default False, Whether to load the current batch into RAM.')
flags.DEFINE_string('out_dir', 'samples', 'Full path to the out directory.')
if __name__ == '__main__':
rbm_layers = utilities.flag_to_list(FLAGS.rbm_layers, 'int')
models_dir = os.path.join(config.models_dir, FLAGS.main_dir)
data_dir = os.path.join(config.data_dir, FLAGS.main_dir)
summary_dir = os.path.join(config.summary_dir, FLAGS.main_dir)
# if FLAGS.finetuned:
rbm_layers = [FLAGS.num_visible] + rbm_layers
model = DBNFinetuned(layers=rbm_layers,
visible_unit_type=FLAGS.visible_unit_type,
hidden_unit_type=FLAGS.hidden_unit_type,
model_name=FLAGS.model_name,
main_dir=FLAGS.main_dir,
models_dir=models_dir,
data_dir=data_dir,
summary_dir=summary_dir)
def do_predict(FLAGS, predictions_dir, num):
dataset_path = "/media/files/yp/rbm/yinchuansanqu/dataset/"
f_list1 = os.listdir(dataset_path)
for i in f_list1:
if os.path.splitext(i)[1] == '.npy':
# 将自动编码器层参数从字符串转换为其特定类型
dae_layers = utilities.flag_to_list(FLAGS.dae_layers, 'int')
dae_enc_act_func = utilities.flag_to_list(FLAGS.dae_enc_act_func, 'str')
dae_dec_act_func = utilities.flag_to_list(FLAGS.dae_dec_act_func, 'str')
dae_opt = utilities.flag_to_list(FLAGS.dae_opt, 'str')
dae_loss_func = utilities.flag_to_list(FLAGS.dae_loss_func, 'str')
dae_learning_rate = utilities.flag_to_list(FLAGS.dae_learning_rate, 'float')
dae_regcoef = utilities.flag_to_list(FLAGS.dae_regcoef, 'float')
dae_corr_type = utilities.flag_to_list(FLAGS.dae_corr_type, 'str')
dae_corr_frac = utilities.flag_to_list(FLAGS.dae_corr_frac, 'float')
dae_num_epochs = utilities.flag_to_list(FLAGS.dae_num_epochs, 'int')
dae_batch_size = utilities.flag_to_list(FLAGS.dae_batch_size, 'int')
# 检查参数
assert all([0. <= cf <= 1. for cf in dae_corr_frac])
assert all([ct in ['masking', 'salt_and_pepper', 'none'] for ct in dae_corr_type])
assert FLAGS.dataset in ['mnist', 'cifar10', 'custom']
assert len(dae_layers) > 0
assert all([af in ['sigmoid', 'tanh'] for af in dae_enc_act_func])
assert all([af in ['sigmoid', 'tanh', 'none'] for af in dae_dec_act_func])
utilities.random_seed_np_tf(FLAGS.seed)
def load_from_np(dataset_path):
if dataset_path != '':
return np.load(dataset_path)
else:
return None
# 创建编码、解码、微调函数和网络模型对象
sdae = None
dae_enc_act_func = [utilities.str2actfunc(af) for af in dae_enc_act_func]
dae_dec_act_func = [utilities.str2actfunc(af) for af in dae_dec_act_func]
finetune_act_func = utilities.str2actfunc(FLAGS.finetune_act_func)
sdae = stacked_denoising_autoencoder.StackedDenoisingAutoencoder(
do_pretrain=FLAGS.do_pretrain, name=FLAGS.name,
layers=dae_layers, finetune_loss_func=FLAGS.finetune_loss_func,
finetune_learning_rate=FLAGS.finetune_learning_rate, finetune_num_epochs=FLAGS.finetune_num_epochs,
finetune_opt=FLAGS.finetune_opt, finetune_batch_size=FLAGS.finetune_batch_size,
finetune_dropout=FLAGS.finetune_dropout,
enc_act_func=dae_enc_act_func, dec_act_func=dae_dec_act_func,
corr_type=dae_corr_type, corr_frac=dae_corr_frac, regcoef=dae_regcoef,
loss_func=dae_loss_func, opt=dae_opt,
learning_rate=dae_learning_rate, momentum=FLAGS.momentum,
num_epochs=dae_num_epochs, batch_size=dae_batch_size,
finetune_act_func=finetune_act_func)
# 训练模型 (无监督预训练)
if FLAGS.do_pretrain:
encoded_X, encoded_vX = sdae.pretrain(trX, vlX)
FLAGS.test_dataset = dataset_path + i
# FLAGS.test_labels = "/media/files/yp/rbm/pic_div/label/binary/" + str(idx) + ".npy"
FLAGS.save_predictions = predictions_dir + i
# teX, teY = load_from_np(FLAGS.test_dataset), load_from_np(FLAGS.test_labels)
teX = load_from_np(FLAGS.test_dataset)
# 计算模型在测试集上的准确率
# print('Test set accuracy: {}'.format(sdae.score(teX, teY)))
# 保存模型的预测
if FLAGS.save_predictions:
print('Saving the predictions for the test set...')
predict = sdae.predict(teX).astype(np.float16)
np.save(FLAGS.save_predictions, predict)
# for idx in range(0, num):
# # 将自动编码器层参数从字符串转换为其特定类型
# dae_layers = utilities.flag_to_list(FLAGS.dae_layers, 'int')
# dae_enc_act_func = utilities.flag_to_list(FLAGS.dae_enc_act_func, 'str')
# dae_dec_act_func = utilities.flag_to_list(FLAGS.dae_dec_act_func, 'str')
# dae_opt = utilities.flag_to_list(FLAGS.dae_opt, 'str')
# dae_loss_func = utilities.flag_to_list(FLAGS.dae_loss_func, 'str')
# dae_learning_rate = utilities.flag_to_list(FLAGS.dae_learning_rate, 'float')
# dae_regcoef = utilities.flag_to_list(FLAGS.dae_regcoef, 'float')
# dae_corr_type = utilities.flag_to_list(FLAGS.dae_corr_type, 'str')
# dae_corr_frac = utilities.flag_to_list(FLAGS.dae_corr_frac, 'float')
# dae_num_epochs = utilities.flag_to_list(FLAGS.dae_num_epochs, 'int')
# dae_batch_size = utilities.flag_to_list(FLAGS.dae_batch_size, 'int')
#
# # 检查参数
# assert all([0. <= cf <= 1. for cf in dae_corr_frac])
# assert all([ct in ['masking', 'salt_and_pepper', 'none'] for ct in dae_corr_type])
# assert FLAGS.dataset in ['mnist', 'cifar10', 'custom']
# assert len(dae_layers) > 0
# assert all([af in ['sigmoid', 'tanh'] for af in dae_enc_act_func])
# assert all([af in ['sigmoid', 'tanh', 'none'] for af in dae_dec_act_func])
#
# utilities.random_seed_np_tf(FLAGS.seed)
#
# def load_from_np(dataset_path):
# if dataset_path != '':
# return np.load(dataset_path)
# else:
# return None
#
# # 创建编码、解码、微调函数和网络模型对象
# sdae = None
#
# dae_enc_act_func = [utilities.str2actfunc(af) for af in dae_enc_act_func]
# dae_dec_act_func = [utilities.str2actfunc(af) for af in dae_dec_act_func]
# finetune_act_func = utilities.str2actfunc(FLAGS.finetune_act_func)
#
# sdae = stacked_denoising_autoencoder.StackedDenoisingAutoencoder(
# do_pretrain=FLAGS.do_pretrain, name=FLAGS.name,
# layers=dae_layers, finetune_loss_func=FLAGS.finetune_loss_func,
# finetune_learning_rate=FLAGS.finetune_learning_rate, finetune_num_epochs=FLAGS.finetune_num_epochs,
# finetune_opt=FLAGS.finetune_opt, finetune_batch_size=FLAGS.finetune_batch_size,
# finetune_dropout=FLAGS.finetune_dropout,
# enc_act_func=dae_enc_act_func, dec_act_func=dae_dec_act_func,
# corr_type=dae_corr_type, corr_frac=dae_corr_frac, regcoef=dae_regcoef,
# loss_func=dae_loss_func, opt=dae_opt,
# learning_rate=dae_learning_rate, momentum=FLAGS.momentum,
# num_epochs=dae_num_epochs, batch_size=dae_batch_size,
# finetune_act_func=finetune_act_func)
#
# # 训练模型 (无监督预训练)
# if FLAGS.do_pretrain:
# encoded_X, encoded_vX = sdae.pretrain(trX, vlX)
#
# FLAGS.test_dataset = "/media/files/yp/rbm/pic_div/dataset/test" + str(idx) + "_RGB.npy"
# # FLAGS.test_labels = "/media/files/yp/rbm/pic_div/label/binary/" + str(idx) + ".npy"
# FLAGS.save_predictions = predictions_dir + str(idx) + ".npy"
# # teX, teY = load_from_np(FLAGS.test_dataset), load_from_np(FLAGS.test_labels)
# teX = load_from_np(FLAGS.test_dataset)
# # 计算模型在测试集上的准确率
# # print('Test set accuracy: {}'.format(sdae.score(teX, teY)))
#
# # 保存模型的预测
# if FLAGS.save_predictions:
# print('Saving the predictions for the test set...')
# np.save(FLAGS.save_predictions, sdae.predict(teX))
#
# def save_layers_output(which_set):
#
# if which_set == 'test':
# teout = sdae.get_layers_output(teX)
# for i, o in enumerate(teout):
# np.save(FLAGS.save_layers_output_test + '-layer-' + str(i + 1) + '-test', o)
#
# # 保存模型每一层对测试集的输出
# if FLAGS.save_layers_output_test:
# print('Saving the output of each layer for the test set')
# save_layers_output('test')
#
# # 保存模型每一层对训练集的输出
# if FLAGS.save_layers_output_train:
# print('Saving the output of each layer for the train set')
# save_layers_output('train')
#
print '-------------------------------------------预测过程已经完成---------------------------------------------------'
flags.DEFINE_string('rbm_learning_rate', '0.01,', 'Initial learning rate.')
flags.DEFINE_string('rbm_num_epochs', '10,', 'Number of epochs.')
flags.DEFINE_string('rbm_batch_size', '10,', 'Size of each mini-batch.')
flags.DEFINE_string('rbm_gibbs_k', '1,', 'Gibbs sampling steps.')
# Supervised fine tuning parameters
flags.DEFINE_float('finetune_learning_rate', 0.01, 'Learning rate.')
flags.DEFINE_string('finetune_enc_act_func', 'relu,', 'Activation function for the encoder fine-tuning phase. ["sigmoid, "tanh", "relu"]')
flags.DEFINE_string('finetune_dec_act_func', 'sigmoid,', 'Activation function for the decoder fine-tuning phase. ["sigmoid, "tanh", "relu"]')
flags.DEFINE_integer('finetune_num_epochs', 10, 'Number of epochs.')
flags.DEFINE_integer('finetune_batch_size', 10, 'Size of each mini-batch.')
flags.DEFINE_string('finetune_opt', 'gradient_descent', '["gradient_descent", "ada_grad", "momentum", "adam"]')
flags.DEFINE_string('finetune_loss_func', 'mean_squared', 'Loss function.')
flags.DEFINE_float('finetune_dropout', 1, 'Dropout parameter.')
# Conversion of Autoencoder layers parameters from string to their specific type
rbm_names = utilities.flag_to_list(FLAGS.rbm_names, 'str')
rbm_layers = utilities.flag_to_list(FLAGS.rbm_layers, 'int')
rbm_learning_rate = utilities.flag_to_list(FLAGS.rbm_learning_rate, 'float')
rbm_num_epochs = utilities.flag_to_list(FLAGS.rbm_num_epochs, 'int')
rbm_batch_size = utilities.flag_to_list(FLAGS.rbm_batch_size, 'int')
rbm_gibbs_k = utilities.flag_to_list(FLAGS.rbm_gibbs_k, 'int')
finetune_enc_act_func = utilities.flag_to_list(FLAGS.finetune_enc_act_func, 'str')
finetune_dec_act_func = utilities.flag_to_list(FLAGS.finetune_dec_act_func, 'str')
# Parameters validation
assert FLAGS.dataset in ['mnist', 'cifar10', 'custom']
assert FLAGS.finetune_loss_func in ["mean_squared", "cross_entropy"]
assert len(rbm_layers) > 0
if __name__ == '__main__':
'Activation function for the decoder. ["sigmoid", "tanh", "none"]')
flags.DEFINE_string('dae_loss_func', 'mse,',
'Loss function. ["mse" or "cross_entropy"]')
flags.DEFINE_string('dae_opt', 'sgd,',
'["sgd", "ada_grad", "momentum", "adam"]')
flags.DEFINE_string('dae_learning_rate', '0.01,', 'Initial learning rate.')
flags.DEFINE_string('dae_num_epochs', '10,', 'Number of epochs.')
flags.DEFINE_string('dae_batch_size', '10,', 'Size of each mini-batch.')
flags.DEFINE_string(
'dae_corr_type', 'none,',
'Type of input corruption. ["none", "masking", "salt_and_pepper"]')
flags.DEFINE_string('dae_corr_frac', '0.0,',
'Fraction of the input to corrupt.')
# Conversion of Autoencoder layers parameters from string to their specific type
dae_layers = utilities.flag_to_list(FLAGS.dae_layers, 'int')
dae_enc_act_func = utilities.flag_to_list(FLAGS.dae_enc_act_func, 'str')
dae_dec_act_func = utilities.flag_to_list(FLAGS.dae_dec_act_func, 'str')
dae_opt = utilities.flag_to_list(FLAGS.dae_opt, 'str')
dae_loss_func = utilities.flag_to_list(FLAGS.dae_loss_func, 'str')
dae_learning_rate = utilities.flag_to_list(FLAGS.dae_learning_rate, 'float')
dae_regcoef = utilities.flag_to_list(FLAGS.dae_regcoef, 'float')
dae_corr_type = utilities.flag_to_list(FLAGS.dae_corr_type, 'str')
dae_corr_frac = utilities.flag_to_list(FLAGS.dae_corr_frac, 'float')
dae_num_epochs = utilities.flag_to_list(FLAGS.dae_num_epochs, 'int')
dae_batch_size = utilities.flag_to_list(FLAGS.dae_batch_size, 'int')
finetune_enc_act_func = utilities.flag_to_list(FLAGS.finetune_enc_act_func,
'str')
finetune_dec_act_func = utilities.flag_to_list(FLAGS.finetune_dec_act_func,
'str')
def do_predict(internal_FLAGS, internal_dataset):
# 将自动编码器层参数从字符串转换为其特定类型
dae_layers = utilities.flag_to_list(internal_FLAGS.dae_layers, 'int')
dae_enc_act_func = utilities.flag_to_list(internal_FLAGS.dae_enc_act_func,
'str')
dae_dec_act_func = utilities.flag_to_list(internal_FLAGS.dae_dec_act_func,
'str')
dae_opt = utilities.flag_to_list(internal_FLAGS.dae_opt, 'str')
dae_loss_func = utilities.flag_to_list(internal_FLAGS.dae_loss_func, 'str')
dae_learning_rate = utilities.flag_to_list(
internal_FLAGS.dae_learning_rate, 'float')
dae_regcoef = utilities.flag_to_list(internal_FLAGS.dae_regcoef, 'float')
dae_corr_type = utilities.flag_to_list(internal_FLAGS.dae_corr_type, 'str')
dae_corr_frac = utilities.flag_to_list(internal_FLAGS.dae_corr_frac,
'float')
dae_num_epochs = utilities.flag_to_list(internal_FLAGS.dae_num_epochs,
'int')
dae_batch_size = utilities.flag_to_list(internal_FLAGS.dae_batch_size,
'int')
# 检查参数
assert all([0. <= cf <= 1. for cf in dae_corr_frac])
assert all(
[ct in ['masking', 'salt_and_pepper', 'none'] for ct in dae_corr_type])
assert internal_FLAGS.dataset in ['mnist', 'cifar10', 'custom']
assert len(dae_layers) > 0
assert all([af in ['sigmoid', 'tanh'] for af in dae_enc_act_func])
assert all([af in ['sigmoid', 'tanh', 'none'] for af in dae_dec_act_func])
utilities.random_seed_np_tf(internal_FLAGS.seed)
# 创建编码、解码、微调函数和网络模型对象
sdae = None
dae_enc_act_func = [utilities.str2actfunc(af) for af in dae_enc_act_func]
dae_dec_act_func = [utilities.str2actfunc(af) for af in dae_dec_act_func]
finetune_act_func = utilities.str2actfunc(internal_FLAGS.finetune_act_func)
sdae = stacked_denoising_autoencoder.StackedDenoisingAutoencoder(
do_pretrain=internal_FLAGS.do_pretrain,
name=internal_FLAGS.name,
layers=dae_layers,
finetune_loss_func=internal_FLAGS.finetune_loss_func,
finetune_learning_rate=internal_FLAGS.finetune_learning_rate,
finetune_num_epochs=internal_FLAGS.finetune_num_epochs,
finetune_opt=internal_FLAGS.finetune_opt,
finetune_batch_size=internal_FLAGS.finetune_batch_size,
finetune_dropout=internal_FLAGS.finetune_dropout,
enc_act_func=dae_enc_act_func,
dec_act_func=dae_dec_act_func,
corr_type=dae_corr_type,
corr_frac=dae_corr_frac,
regcoef=dae_regcoef,
loss_func=dae_loss_func,
opt=dae_opt,
learning_rate=dae_learning_rate,
momentum=internal_FLAGS.momentum,
num_epochs=dae_num_epochs,
batch_size=dae_batch_size,
finetune_act_func=finetune_act_func)
# 训练模型 (无监督预训练)
if internal_FLAGS.do_pretrain:
encoded_X, encoded_vX = sdae.pretrain(trX, vlX)
teX = internal_dataset
print('Saving the predictions for the test set...')
internal_predictions = sdae.predict(teX)
return internal_predictions
'Comma-separated values for the number of hidden units in the layers.')
flags.DEFINE_string('datafiles', 'filename_missing,',
'Comma-separated names of the input files.')
flags.DEFINE_string('rbm_batch_sizes', '10,', 'Size of each mini-batch.')
flags.DEFINE_boolean(
'memory_mapping', True,
'default True, Whether to use memory mapping to avoid loading the full dataset into RAM.'
)
flags.DEFINE_boolean(
'load_batch', False,
'default False, Whether to load the current batch into RAM.')
flags.DEFINE_boolean('finetuned', False,
'default False, Whether to load a finetuned model.')
if __name__ == '__main__':
rbm_layers = utilities.flag_to_list(FLAGS.rbm_layers, 'int')
datafiles = utilities.flag_to_list(FLAGS.datafiles, 'str')
models_dir = os.path.join(config.models_dir, FLAGS.main_dir)
data_dir = os.path.join(config.data_dir, FLAGS.main_dir)
summary_dir = os.path.join(config.summary_dir, FLAGS.main_dir)
if FLAGS.finetuned:
rbm_layers = [FLAGS.num_visible] + rbm_layers
model = DBNFinetuned(layers=rbm_layers,
visible_unit_type=FLAGS.visible_unit_type,
hidden_unit_type=FLAGS.hidden_unit_type,
model_name=FLAGS.model_name,
main_dir=FLAGS.main_dir,
models_dir=models_dir,
data_dir=data_dir,
def do_predict(FLAGS, datasets):
"""
调用堆栈降噪自动编码器进行预测
:param FLAGS: tensorflow.app.flags.FLAGS 类型。包含了定义的网络模型的各项参数
:param datasets: ndarray 类型。制作好的测试数据
:return: ndarray 类型。预测结果
"""
# 将自动编码器层参数从字符串转换为其特定类型
dae_layers = utilities.flag_to_list(FLAGS.dae_layers, 'int')
dae_enc_act_func = utilities.flag_to_list(FLAGS.dae_enc_act_func, 'str')
dae_dec_act_func = utilities.flag_to_list(FLAGS.dae_dec_act_func, 'str')
dae_opt = utilities.flag_to_list(FLAGS.dae_opt, 'str')
dae_loss_func = utilities.flag_to_list(FLAGS.dae_loss_func, 'str')
dae_learning_rate = utilities.flag_to_list(FLAGS.dae_learning_rate, 'float')
dae_regcoef = utilities.flag_to_list(FLAGS.dae_regcoef, 'float')
dae_corr_type = utilities.flag_to_list(FLAGS.dae_corr_type, 'str')
dae_corr_frac = utilities.flag_to_list(FLAGS.dae_corr_frac, 'float')
dae_num_epochs = utilities.flag_to_list(FLAGS.dae_num_epochs, 'int')
dae_batch_size = utilities.flag_to_list(FLAGS.dae_batch_size, 'int')
# 检查参数
assert all([0. <= cf <= 1. for cf in dae_corr_frac])
assert all([ct in ['masking', 'salt_and_pepper', 'none'] for ct in dae_corr_type])
assert FLAGS.dataset in ['mnist', 'cifar10', 'custom']
assert len(dae_layers) > 0
assert all([af in ['sigmoid', 'tanh'] for af in dae_enc_act_func])
assert all([af in ['sigmoid', 'tanh', 'none'] for af in dae_dec_act_func])
utilities.random_seed_np_tf(FLAGS.seed)
# 创建编码、解码、微调函数和网络模型对象
dae_enc_act_func = [utilities.str2actfunc(af) for af in dae_enc_act_func]
dae_dec_act_func = [utilities.str2actfunc(af) for af in dae_dec_act_func]
finetune_act_func = utilities.str2actfunc(FLAGS.finetune_act_func)
sdae = stacked_denoising_autoencoder.StackedDenoisingAutoencoder(
do_pretrain=FLAGS.do_pretrain, name=FLAGS.name,
layers=dae_layers, finetune_loss_func=FLAGS.finetune_loss_func,
finetune_learning_rate=FLAGS.finetune_learning_rate,
finetune_num_epochs=FLAGS.finetune_num_epochs,
finetune_opt=FLAGS.finetune_opt, finetune_batch_size=FLAGS.finetune_batch_size,
finetune_dropout=FLAGS.finetune_dropout,
enc_act_func=dae_enc_act_func, dec_act_func=dae_dec_act_func,
corr_type=dae_corr_type, corr_frac=dae_corr_frac, regcoef=dae_regcoef,
loss_func=dae_loss_func, opt=dae_opt,
learning_rate=dae_learning_rate, momentum=FLAGS.momentum,
num_epochs=dae_num_epochs, batch_size=dae_batch_size,
finetune_act_func=finetune_act_func)
# 训练模型 (无监督预训练)
# if FLAGS.do_pretrain:
# encoded_X, encoded_vX = sdae.pretrain(trX, vlX)
teX = datasets
# print('Saving the predictions for the test set...')
internal_predictions = sdae.predict(teX)
return internal_predictions
