미카 프로젝트

optimizer = tf.train.AdamOptimizer(learning_rate)

train = optimizer.minimize(cost)

saver = tf.train.Saver()

sess = tf.Session()

sess.run(tf.global_variables_initializer())

# training 을 저장할 file 명

mytrain= "./mytain.ckpt"

if os.path.exists(mytrain+".meta"):

# 파일에서 loading

    saver.restore(sess, mytrain)

else :

    for step in range(2001) :

        cost_val, hy_val, _ = sess.run([cost, hypothesis, train], feed_dict={X: trainX, Y: trainY})

        if step % 10 ==0 :

            print(step, "Cost: ", cost_val, "\nPrediction:\n", hy_val)

#file에 저장

    saver.save(sess, mytrain)

test_predict = sess.run(hypothesis, feed_dict= {X: testX})

train_size = int(len(dataY) * 0.7)

test_size =  len(dataY) - train_size

trainX = np.array(dataX[0:train_size])

testX = np.array(dataX[train_size:len(dataX)])

trainY = np.array(dataY[0:train_size])

testY = np.array(dataY[train_size:len(dataY)])

input_len = data_dim*seq_length

X = tf.placeholder(tf.float32, [None, input_len], name='X')

Y = tf.placeholder(tf.float32, [None, 1], name='Y')

W = tf.Variable(tf.random_normal([input_len, 1]), name='weight')

b = tf.Variable(tf.random_normal([1]), name='bias')

#hypothesis = tf.matmul(X, W) + b

hypothesis = tf.add(b, tf.matmul(X, W), name="h")

cost = tf.reduce_mean(tf.square(hypothesis - Y))

optimizer = tf.train.AdamOptimizer(learning_rate)

train = optimizer.minimize(cost)

saver = tf.train.Saver()

sess = tf.Session()

sess.run(tf.global_variables_initializer())

for step in range(2001) :

    cost_val, hy_val, _ = sess.run([cost, hypothesis, train], feed_dict={X: trainX, Y: trainY})

    if step % 10 ==0 :

        print(step, "Cost: ", cost_val, "\nPrediction:\n", hy_val)

saver.save(sess, mytrain)

if os.path.exists(mytrain+".meta"):

    sess = tf.Session()

    sess.run(tf.global_variables_initializer())

    saver = tf.train.import_meta_graph(mytrain+".meta")

    saver.restore(sess, mytrain)

    graph = tf.get_default_graph()

    X = graph.get_tensor_by_name('X:0')

    W = graph.get_tensor_by_name('weight:0')

    b = graph.get_tensor_by_name('bias:0')

    hypothesis = graph.get_tensor_by_name('h:0')   #tf.matmul(X, W) + b

    test_predict = sess.run(hypothesis, feed_dict={X: testX})

import matplotlib.pyplot as plt

plt.figure()

plt.plot([1,2,3,4,5,6,7,8,9,8,7,6,5,4,3,2,1,0])

plt.show()

print("hello")

import tensorflow as tf

import numpy as np

import matplotlib.pyplot as plt

sess = tf.InteractiveSession()

image = np.array ([[[[1],[2],[3]],

                    [[4],[5],[6]],

                    [[7],[8],[9]]]], dtype=np.float32)

print(image.shape)

# visualization

plt.imshow(image.reshape(3,3), cmap='Greys')

print("image.shape", image.shape)

#weight = tf.constant([[[[1.]],[[1.]]],

#                      [[[1.]],[[1.]]]])

weight = tf.constant([[[[1.,10.,-1]],[[1.,10.,-1]]],

                      [[[1.,10.,-1]],[[1.,10.,-1]]]])

print("weight.shape", weight.shape)

conv2d = tf.nn.conv2d(image, weight, strides =[1,1,1,1], padding='SAME') # convolution - padding ='VALID'

conv2d_img = conv2d.eval()

print("conv2d_img.shape", conv2d_img.shape)

# visualization

conv2d_img = np.swapaxes(conv2d_img, 0,3)

for i , one_img in enumerate(conv2d_img):

        print(one_img.reshape(3,3))

        plt.subplot(1,3,i+1), plt.imshow(one_img.reshape(3,3), cmap='gray')

plt.show()

pool = tf.nn.max_pool(conv2d_img, ksize=[1,2,2,1],

                    strides=[1,1,1,1],padding='SAME')

print("pool", pool)