def main():
num_classes = 10
num_hidden_layers = 1024
(x_train, y_train), (x_test, y_test) = mnist.load_data()
# Process images into input vectors
# each mnist image is a 28x28 picture with value ranges between 0 and 255
x_train = x_train.astype(np.float32) / 255.
x_train = x_train.reshape(-1, 28**2)
x_test = x_test.astype(np.float32) / 255.
x_test = x_test.reshape(-1, 28**2)
# converts [1,2] into [[0,1,0], [0,0,1]]
y_train = to_categorical(y_train, num_classes).astype(np.float32)
y_test = to_categorical(y_test, num_classes).astype(np.float32)
# create instance of our model
model = ELM(28**2, num_hidden_layers, num_classes)
# Train
model.fit(x_train, y_train)
train_loss, train_acc = model.evaluate(x_train, y_train)
print('train loss: %f' % train_loss)
print('train acc: %f' % train_acc)
# Validation
val_loss, val_acc = model.evaluate(x_test, y_test)
print('val loss: %f' % val_loss)
print('val acc: %f' % val_acc)
from model import ELM
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
# Basic tf setting
tf.set_random_seed(2016)
sess = tf.Session()
# Get data
mnist = input_data.read_data_sets("/tmp/data/", one_hot=True)
# Construct ELM
batch_size = 50000
hidden_num = 50000
print("batch_size : {}".format(batch_size))
print("hidden_num : {}".format(hidden_num))
elm = ELM(sess, batch_size, 784, hidden_num, 10)
# one-step feed-forward training
train_x, train_y = mnist.train.next_batch(batch_size)
elm.feed(train_x, train_y)
# testing
elm.test(mnist.test.images, mnist.test.labels)
FLAGS = tf.app.flags.FLAGS
num = FLAGS.num
# Basic tf setting
tf.set_random_seed(2016)
sess = tf.Session()
# Get data
# mnist = input_data.read_data_sets("/tmp/data/", one_hot=True)
# Construct ELM
batch_size = TRAIN_SIZE
hidden_num = 150
print("batch_size : {}".format(batch_size))
print("hidden_num : {}".format(hidden_num))
elm = ELM(sess, batch_size, NUM_ROWS * DATA_SIZE, hidden_num, 4)
BASE = './sensor4_1_driver'
# train_data, train_labels = extractData_oned_train_val.extract_data_oned(numRows=NUM_ROWS, numData=TRAIN_SIZE,
# drivers=DRIVERS, labels=LABELS, mode='train',
# DATA_SIZE=DATA_SIZE,
# NUM_CHANNELS=NUM_CHANNELS, ONED=True)
# train_data=[TRAIN_SIZE,600]
validation_data, validation_labels = extractData_oned_train_val.extract_data_oned(
numRows=NUM_ROWS,
numData=VALIDATION_SIZE,
drivers=DRIVERS,
labels=LABELS,
mode='validate',
DATA_SIZE=DATA_SIZE,
NUM_CHANNELS=NUM_CHANNELS,
ONED=True,
FLAGS = tf.app.flags.FLAGS
num = FLAGS.num
# Basic tf setting
tf.set_random_seed(2016)
sess = tf.Session()
# Get data
# mnist = input_data.read_data_sets("/tmp/data/", one_hot=True)
# Construct ELM
batch_size = TRAIN_SIZE
hidden_num = 150
print("batch_size : {}".format(batch_size))
print("hidden_num : {}".format(hidden_num))
elm = ELM(sess, batch_size, NUM_ROWS * DATA_SIZE, hidden_num, 4)
BASE = './sensor10_driver'
# train_data, train_labels = extractData_oned_train_val.extract_data_oned(numRows=NUM_ROWS, numData=TRAIN_SIZE,
# drivers=DRIVERS, labels=LABELS, mode='train',
# DATA_SIZE=DATA_SIZE,
# NUM_CHANNELS=NUM_CHANNELS, ONED=True)
# train_data=[TRAIN_SIZE,600]
validation_data, validation_labels = extractData_oned_train_val.extract_data_oned(
numRows=NUM_ROWS,
numData=VALIDATION_SIZE,
drivers=DRIVERS,
labels=LABELS,
mode='validate',
DATA_SIZE=DATA_SIZE,
NUM_CHANNELS=NUM_CHANNELS,
ONED=True,