def main(ckpt = None):
#with tf.Graph().as_default():
with tf.Session().graph.as_default():
keep_prob = tf.placeholder("float")
# データ準備
images, labels, _ = data_input.load_data([FLAGS.train], FLAGS.batch_size, shuffle = True, distored = True)
# モデル構築
logits = model.inference_deep(images, keep_prob, data_input.DST_LONG_SIZE,data_input.DST_SHORT_SIZE, data_input.NUM_CLASS)
loss_value = model.loss(logits, labels)
train_op = model.training(loss_value, FLAGS.learning_rate)
acc = model.accuracy(logits, labels)
saver = tf.train.Saver(max_to_keep = 0)
sess = tf.Session()
sess.run(tf.initialize_all_variables())
if ckpt:
print 'restore ckpt', ckpt
saver.restore(sess, ckpt)
tf.train.start_queue_runners(sess)
summary_op = tf.merge_all_summaries()
summary_writer = tf.train.SummaryWriter(FLAGS.train_dir, sess.graph_def)
#モデル構築をモニタリング
for step in range(FLAGS.max_steps):
start_time = time.time()
_, loss_result, acc_res = sess.run([train_op, loss_value, acc], feed_dict={keep_prob: 0.99})
duration = time.time() - start_time
if step % 10 == 0:
num_examples_per_step = FLAGS.batch_size
examples_per_sec = num_examples_per_step / duration
sec_per_batch = float(duration)
format_str = ('%s: step %d, loss = %.2f (%.1f examples/sec; %.3f sec/batch)')
print (format_str % (datetime.now(), step, loss_result, examples_per_sec, sec_per_batch))
print 'acc_res', acc_res
if step % 100 == 0:
summary_str = sess.run(summary_op,feed_dict={keep_prob: 1.0})
summary_writer.add_summary(summary_str, step)
if step % 1000 == 0 or (step + 1) == FLAGS.max_steps or loss_result == 0:
checkpoint_path = os.path.join(FLAGS.train_dir, 'model.ckpt')
save_path = saver.save(sess, checkpoint_path, global_step=step)
print('%s saved' % save_path)
if loss_result == 0:
print('loss is zero')
break
def main():
data = data_input.load_data(TRN_FILE, VAL_FILE, skip_rows=1)
for i in range(ENS_N):
MODEL_DIR = RES_DIR + '/model' + str(i)
METAGRAPH_DIR = MODEL_DIR + '/out'
LOG_DIR = MODEL_DIR + '/log'
for DIR in (MODEL_DIR, LOG_DIR, METAGRAPH_DIR):
try:
os.mkdir(DIR)
except FileExistsError:
pass
print('Architecture: {}'.format(ARCHITECTURE), flush=True)
m = fc.FC(ARCHITECTURE, HYPERPARAMS, log_dir=LOG_DIR,save_graph_def=True)
print('Resetting training epochs', flush=True)
data.train.reset_epochs()
print('Resetting validation epochs', flush=True)
data.validation.reset_epochs()
m.train(data, max_iter=MAX_ITER, max_epochs=MAX_EPOCHS, cross_validate=True,
verbose=True, save=True, save_log=True,outdir=METAGRAPH_DIR)
print("Trained!",flush=True)
del m
tf.reset_default_graph()
import numpy as np
import sys
sys.path.append("../process")
import data_input
from train_network import Lenet
def train(aug, model, train_x, train_y, test_x, test_y):
model.compile(loss="categorical_crossentropy", optimizer="Adam", metrics=["accuracy"])
model.fit(aug.flow(train_x,train_y,batch_size=batch_size), validation_data=(test_x,test_y), steps_per_epoch=len(train_x)//batch_size, epochs=epochs, verbose=1)
model.save("../predict/item_model.h5")
if __name__ =="__main__":
channel = 3
height = 32
width = 32
class_num = 41
norm_size = 32
batch_size = 32
epochs = 40
model = Lenet.neural(channel=channel, height=height, width=width, classes=class_num)
train_x, train_y = data_input.load_data("../data/item_train", norm_size, class_num)
test_x, test_y = data_input.load_data("../data/item_val", norm_size, class_num)
aug = ImageDataGenerator(rotation_range=30, width_shift_range=0.1, height_shift_range=0.1, shear_range=0.2, zoom_range=0.2, horizontal_flip=True, fill_mode="nearest")
train(aug,model,train_x,train_y,test_x,test_y)
from data_input import load_data from helper_files.data3d_augment_reformat import augment_n_reformat # Get hyperparameters hparams = hyperparameters.create_hparams() # Set data input path train_data_dir = hparams.input_dir +'/train' # Set path to save images to during training current_results_dir_base = os.path.join(hparams.results_dir, 'during_training') images_output_path = os.path.join(current_results_dir_base, hparams.model_subdir) # Load preprocessed data images, labels = load_data(train_data_dir) # For the first epoch, do not do data augmentation images_aug, labels_aug = augment_n_reformat(images, labels, hparams) # Set up input data placeholder variables x = tf.placeholder(dtype = tf.float32, shape=images_aug.shape) y = tf.placeholder(dtype = tf.float32, shape=labels_aug.shape) input_images = tf.Variable(x, trainable=False, collections=[]) input_labels = tf.Variable(y, trainable=False, collections=[]) image, label = tf.train.slice_input_producer([input_images, input_labels]) images_batch, labels_batch = tf.train.batch([image, label], batch_size=hparams.batch_size) # Feed input data batch into neural network architecture logits, _ = model(images_batch, hparams) # Calculate loss of current network
def load_data(self):
self.x_train, self.y_train, self.x_valid, self.y_valid, self.x_test, self.y_test, self.question, self.word_embedding, self.max_q_len, self.max_sentences, self.max_sen_len, self.vocab_size, self.question_num = data_input.load_data(
self.config)
self.encoding = _position_encoding(self.max_sen_len,
self.config.embedding_size)
self.train = zip(self.question, self.x_train, self.y_train)
self.valid = zip(self.question, self.x_valid, self.y_valid)
self.test = zip(self.question, self.x_test, self.y_test)
N = EPOCHS
plt.plot(np.arange(0, N), _history.history["loss"], label="train_loss")
plt.plot(np.arange(0, N), _history.history["val_loss"], label="val_loss")
plt.plot(np.arange(0, N), _history.history["accuracy"], label="train_acc")
plt.plot(np.arange(0, N),
_history.history["val_accuracy"],
label="val_acc")
plt.title("loss and accuracy")
plt.xlabel("epoch")
plt.ylabel("loss/acc")
plt.legend(loc="best")
plt.savefig("./result/result.png")
plt.show()
if __name__ == "__main__":
train_x, train_y = data_input.load_data("./data/train", NORM_SIZE,
CLASS_NUM)
test_x, test_y = data_input.load_data("./data/test", NORM_SIZE, CLASS_NUM)
aug = ImageDataGenerator(rotation_range=30,
width_shift_range=0.1,
height_shift_range=0.1,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode="nearest")
train(aug, train_x, train_y, test_x, test_y)
plt.savefig("../result/result.png")
plt.show()
if __name__ == "__main__":
channel = 3
height = 32
width = 32
class_num = 62
norm_size = 32 #参数
batch_size = 32
epochs = 40
model = Lenet.neural(channel=channel,
height=height,
width=width,
classes=class_num) #网络
train_x, train_y = data_input.load_data("../data/train", norm_size,
class_num)
test_x, test_y = data_input.load_data("../data/test", norm_size,
class_num) #生成数据
aug = ImageDataGenerator(rotation_range=30,
width_shift_range=0.1,
height_shift_range=0.1,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode="nearest") #数据增强，生成迭代器
train(aug, model, train_x, train_y, test_x, test_y) #训练
from knn import knn
from neural_network import Network
from bayes import Bayes
from svm_class import Svm
from data_input import load_data, load_data_test, data_to_libsvm, testdata_to_libsvm
from progress_until import Progress
numToTrain = 60000
numToClassfy = 10000
model_num = 5
model_now = 0
results = np.zeros([model_num, numToClassfy], dtype=int)
print("开始读取MNIST数据：")
images, label = load_data()
test_images, test_label = load_data_test()
#libsvm格式数据读取
svm_label, svm_images = data_to_libsvm(images, label)
svm_test_label, svm_test_images = testdata_to_libsvm(test_images, test_label)
#数据二值化
images = np.where(images >= 128, 1, 0)
test_images = np.where(test_images >= 128, 1, 0)
network = Network(784, 2, [200, 10], 10)
network.startLearing(numToTrain, images, label, 1)
results[model_now] = network.test(test_images, test_label, numToClassfy)
model_now += 1
print("SVM 1训练中...")