tesorflow 1

# -*- coding: utf-8 -*-

import numpy as np
import tensorflow as tf
from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelBinarizer
import matplotlib.pylab as plt
from tensorflow.examples.tutorials.mnist import input_data

# 用Tensorflow計算 a = (b + c) * (c + 2)
def test01():
	# 首先，創建一個TensorFlow常量=>2
	const = tf.constant(2.0, name='const')

	# 創建TensorFlow變量b和c
	b = tf.Variable(2.0, name='b')
	c = tf.Variable(1.0, dtype=tf.float32, name='c')
	
	# 創建operation
	d = tf.add(b, c, name='d')
	e = tf.add(c, const, name='e')
	a = tf.multiply(d, e, name='a')
	
	# 1. 定義init operation
	init_op = tf.global_variables_initializer()
	
	# session
	with tf.Session() as sess:
		# 2. 運行init operation
		sess.run(init_op)
		# 計算
		a_out = sess.run(a)
		print("Variable a is {}".format(a_out))


def test02():
	# 首先，創建一個TensorFlow常量=>2
	const = tf.constant(2.0, name='const')

	# 創建TensorFlow變量b和c
	b = tf.placeholder(tf.float32, [None, 1], name='b')
	c = tf.Variable(1.0, dtype=tf.float32, name='c')
	
	# 創建operation
	d = tf.add(b, c, name='d')
	e = tf.add(c, const, name='e')
	a = tf.multiply(d, e, name='a')
	
	# 1. 定義init operation
	init_op = tf.global_variables_initializer()
	
	# session
	with tf.Session() as sess:
		# 2. 運行init operation
		sess.run(init_op)
		# 計算
		#a_out = sess.run(a)
		a_out = sess.run(a, feed_dict={b: np.arange(0, 10)[:, np.newaxis]})
		print("Variable a is {}".format(a_out))


def test03():
	mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)
	# 超參數
	learning_rate = 0.5
	epochs = 10
	batch_size = 100

	# placeholder
	# 輸入圖片爲28 x 28 像素 = 784
	x = tf.placeholder(tf.float32, [None, 784])
	# 輸出爲0-9的one-hot編碼
	y = tf.placeholder(tf.float32, [None, 10])
	
	# hidden layer => w, b
	W1 = tf.Variable(tf.random_normal([784, 300], stddev=0.03), name='W1')
	b1 = tf.Variable(tf.random_normal([300]), name='b1')
	# output layer => w, b
	W2 = tf.Variable(tf.random_normal([300, 10], stddev=0.03), name='W2')
	b2 = tf.Variable(tf.random_normal([10]), name='b2')
	
	# hidden layer
	hidden_out = tf.add(tf.matmul(x, W1), b1)
	hidden_out = tf.nn.relu(hidden_out)
	
	# 計算輸出
	out = tf.add(tf.matmul(hidden_out, W2), b2)
	y_ = tf.nn.softmax(out)
	
	# tf.clip_by_value(A, min, max)：輸入一個張量A，把A中的每一個元素的值都壓縮在
	# min和max之間。小於min的讓它等於min，大於max的元素的值等於max。
	y_clipped = tf.clip_by_value(y_, 1e-10, 0.9999999)
	cross_entropy = -tf.reduce_mean(tf.reduce_sum(y * tf.log(y_clipped) + (1 - y) * tf.log(1 - y_clipped), axis=1))
	
	# 創建優化器，確定優化目標
	optimizer = tf.train.GradientDescentOptimizer(learning_rate=learning_rate).minimize(cross_entropy)
	
	# init operator
	init_op = tf.global_variables_initializer()

	# 創建準確率節點
	# 0：按列計算，1：行計算；0：按列計算，1：行計算
	# correct_predicion會返回一個m×1 m\times 1m×1的tensor，tensor的值爲True/False表示是否正確預測。
	correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
	accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

	# 創建session
	with tf.Session() as sess:
		# 變量初始化
		sess.run(init_op)
		total_batch = int(len(mnist.train.labels) / batch_size)
		for epoch in range(epochs):
			avg_cost = 0
			for i in range(total_batch):
				batch_x, batch_y = mnist.train.next_batch(batch_size=batch_size)
				_, c = sess.run([optimizer, cross_entropy], feed_dict={x: batch_x, y: batch_y})
				avg_cost += c / total_batch
			print("Epoch:", (epoch + 1), "cost = ", "{:.3f}".format(avg_cost))
			
		# test	
		print(sess.run(accuracy, feed_dict={x: mnist.test.images, y: mnist.test.labels}))


# 線性迴歸
def test04():
	#create data
	x_data=np.random.rand(100).astype(np.float32)
	y_data=x_data*0.1+0.3

	#create tensorflow structure
	Weights=tf.Variable(tf.random_uniform([1],-1.0,1.0)) #一維，範圍[-1,1]
	biases=tf.Variable(tf.zeros([1]))

	y=Weights*x_data+biases

	# 均方誤差
	loss=tf.reduce_mean(tf.square(y-y_data))

	#建立優化器，減小誤差，提高參數準確度，每次迭代都會優化
	optimizer=tf.train.GradientDescentOptimizer(0.5) #學習效率<1
	train=optimizer.minimize(loss)

	#初始化變量
	init=tf.global_variables_initializer()

	with tf.Session() as sess:
		sess.run(init)
		#train
		for step in range(201):
			sess.run(train)
			if step%20==0:
				print(step,sess.run(Weights),sess.run(biases))


def test05():
	matrix1 = tf.constant([[3, 3]])
	matrix2 = tf.constant([[2], [2]])

	# matrix multiply
	# np.dot(m1,m2)
	product = tf.matmul(matrix1, matrix2)

	# # method 1
	# sess = tf.Session()  # Session是一個object，首字母要大寫
	# # 只有sess.run()之後，tensorflow纔會執行一次
	# result = sess.run(product)
	# print(result)
	# # close 不影響，會顯得更整潔
	# sess.close()

	# method 2
	# with 可以自己關閉會話
	with tf.Session() as sess:
		result2 = sess.run(product)
		print(result2)


def test06():
	state=tf.Variable(0,name='counter')
	# print(state.name)
	# 變量+常量=變量
	one=tf.constant(1)
	new_value=tf.add(state,one)
	
	#將state用new_value代替
	updata=tf.assign(state,new_value)

	#變量必須要激活
	init=tf.global_variables_initializer()

	with tf.Session() as sess:
		sess.run(init)
		for _ in range(3):
			sess.run(updata)
			print(sess.run(state))


# placeholder
def test07():
	# 給定type，tf大部分只能處理float32數據
	input1 = tf.placeholder(tf.float32)
	input2 = tf.placeholder(tf.float32)

	# Tensorflow 1.0 修改版
	# tf.mul---tf.multiply
	# tf.sub---tf.subtract
	# tf.neg---tf.negative
	output = tf.multiply(input1, input2)

	with tf.Session() as sess:
		# placeholder在sess.run()的時候傳入值
		print(sess.run(output, feed_dict={input1: [7.], input2: [2.]}))
		print(sess.run(output, feed_dict={input1: [7., 2], input2: [[2.], [2]]}))


def add_layer(inputs, in_size, out_size, activation_function=None):
    #Weights是一個矩陣，[行，列]爲[in_size,out_size]
    Weights=tf.Variable(tf.random_normal([in_size,out_size]))#正態分佈
	
    #初始值推薦不爲0，所以加上0.1，一行，out_size列
    biases=tf.Variable(tf.zeros([1,out_size])+0.1)
	
    #Weights*x+b的初始化的值，也就是未激活的值
    Wx_plus_b=tf.matmul(inputs,Weights)+biases

    #激活

    if activation_function is None:
        #激活函數爲None，也就是線性函數
        outputs=Wx_plus_b
    else:
        outputs=activation_function(Wx_plus_b)
    return outputs


def test08():
	"""定義數據形式"""
	# (-1,1)之間，有300個單位，後面的是維度，x_data是有300行（300個例子）
	x_data=np.linspace(-1,1,300)[:,np.newaxis]
	
	# 加噪聲,均值爲0，方差爲0.05，大小和x_data一樣
	noise=np.random.normal(0,0.05,x_data.shape)
	y_data=np.square(x_data)-0.5+noise

	xs=tf.placeholder(tf.float32,[None,1])
	ys=tf.placeholder(tf.float32,[None,1])

	"""建立網絡"""
	#定義隱藏層，輸入1個節點，輸出10個節點
	l1=add_layer(xs,1,10,activation_function=tf.nn.relu)
	#定義輸出層
	prediction=add_layer(l1,10,1,activation_function=None)

	"""預測"""
	#損失函數,算出的是每個例子的平方，要求和（reduction_indices=[1]，按行求和）,再求均值
	loss=tf.reduce_mean(tf.reduce_sum(tf.square(ys-prediction),reduction_indices=[1]))

	"""訓練"""
	#優化算法,minimize(loss)以0.1的學習率對loss進行減小
	train_step=tf.train.GradientDescentOptimizer(0.08).minimize(loss)

	init=tf.global_variables_initializer()

	with tf.Session() as sess:
		sess.run(init)
		for i in range(5000):
			sess.run(train_step,feed_dict={xs:x_data,ys:y_data})
			if i%100==0:
				print(sess.run(loss,feed_dict={xs:x_data,ys:y_data}))


def test09():
	"""定義數據形式"""
	# (-1,1)之間，有300個單位，後面的是維度，x_data是有300行（300個例子）
	x_data=np.linspace(-1,1,300)[:,np.newaxis]
	
	# 加噪聲,均值爲0，方差爲0.05，大小和x_data一樣
	noise=np.random.normal(0,0.05,x_data.shape)
	y_data=np.square(x_data)-0.5+noise

	xs=tf.placeholder(tf.float32,[None,1])
	ys=tf.placeholder(tf.float32,[None,1])

	"""建立網絡"""
	#定義隱藏層，輸入1個節點，輸出10個節點
	l1=add_layer(xs,1,10,activation_function=tf.nn.relu)
	#定義輸出層
	prediction=add_layer(l1,10,1,activation_function=None)

	"""預測"""
	#損失函數,算出的是每個例子的平方，要求和（reduction_indices=[1]，按行求和）,再求均值
	loss=tf.reduce_mean(tf.reduce_sum(tf.square(ys-prediction),reduction_indices=[1]))

	"""訓練"""
	#優化算法,minimize(loss)以0.1的學習率對loss進行減小
	train_step=tf.train.GradientDescentOptimizer(0.08).minimize(loss)

	init=tf.global_variables_initializer()

	with tf.Session() as sess:
		sess.run(init)
		
		fig=plt.figure()
		#連續性的畫圖
		ax=fig.add_subplot(1,1,1)
		ax.scatter(x_data,y_data)
		# 不暫停
		plt.ion()
		# plt.show()繪製一次就會暫停
		# plt.show() #也可以用plt.show(block=False)來取消暫停，但是python3.5以後提供了ion的功能，更方便
			
		for i in range(5000):
			sess.run(train_step,feed_dict={xs:x_data,ys:y_data})
			if i%100==0:
				print(sess.run(loss,feed_dict={xs:x_data,ys:y_data}))
				try:
					# 畫出一條後抹除掉,去除第一個線段，但是隻有一個，也就是抹除當前的線段
					ax.lines.remove(lines[0])
				except Exception:
					pass
				
				prediction_value=sess.run(prediction,feed_dict={xs:x_data})
				lines=ax.plot(x_data,prediction_value,'r-',lw=5) #lw線寬

				# 暫停0.1s
				plt.pause(0.1)	
				

# 手寫體
def test10():
	mnist=input_data.read_data_sets('MNIST_data',one_hot=True)
	
	# define placeholder for inputs to networks
	# 不規定有多少個sample，但是每個sample大小爲784（28*28）
	xs=tf.placeholder(tf.float32,[None,784])
	ys=tf.placeholder(tf.float32,[None,10])

	#add output layer
	prediction=add_layer(xs,784,10,activation_function=tf.nn.softmax)

	#the error between prediction and real data
	cross_entropy=tf.reduce_mean(-tf.reduce_sum(ys*tf.log(prediction),reduction_indices=[1]))
	train_strp=tf.train.GradientDescentOptimizer(0.3).minimize(cross_entropy)

	init=tf.global_variables_initializer()
	
	with tf.Session() as sess:
		sess.run(init)
		for i in range(3000):
			batch_xs, batch_ys=mnist.train.next_batch(100)
			sess.run(train_strp,feed_dict={xs:batch_xs,ys:batch_ys})

			if i%100==0:
				y_pre=sess.run(prediction,feed_dict={xs:mnist.test.images})
				correct_prediction=tf.equal(tf.arg_max(y_pre,1),tf.arg_max(mnist.test.labels,1))
				accuracy=tf.reduce_mean(tf.cast(correct_prediction,tf.float32))
				result=sess.run(accuracy,feed_dict={xs:mnist.test.images,ys:mnist.test.labels})
				print(i, " accuracy:", result)


def add_layer_EX(inputs,in_size,out_size,layer_name,activation_function=None):
    #Weights是一個矩陣，[行，列]爲[in_size,out_size]
    Weights=tf.Variable(tf.random_normal([in_size,out_size]))#正態分佈
    #初始值推薦不爲0，所以加上0.1，一行，out_size列
    biases=tf.Variable(tf.zeros([1,out_size])+0.1)
    #Weights*x+b的初始化的值，也就是未激活的值
    Wx_plus_b=tf.matmul(inputs,Weights)+biases

    #激活

    if activation_function is None:
        #激活函數爲None，也就是線性函數
        outputs=Wx_plus_b
    else:
        outputs=activation_function(Wx_plus_b)
    # 下面的表示outputs的值
    tf.summary.histogram(layer_name+'/outputs',outputs)

    return outputs


# 未使用dropout：
def test11():
	#load data
	digits=load_digits()
	#0~9的圖像
	X=digits.data
	print(type(X))
	print("X.shape:", X.shape)
	
	#y是binary的，表示數字1，就在第二個位置放上1，其餘都爲0
	y=digits.target
	print(type(y))
	print("y.shape:", y.shape)
	y=LabelBinarizer().fit_transform(y)
	print(type(y))
	print("y.shape:", y.shape)
	
	#切分
	X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=round(X.shape[0]*0.2))
	print("X_train.shape:", X_train.shape)
	print("y_train.shape:", y_train.shape)
	print("X_test.shape:", X_test.shape)
	print("y_test.shape:", y_test.shape)
	
	#define placeholder for inputs to network
	"""dropout"""
	# 確定保留多少結果不被捨棄掉
	keep_prob=tf.placeholder(tf.float32)
	
	#define placeholder for inputs to network
	xs=tf.placeholder(tf.float32,[None,64])
	ys=tf.placeholder(tf.float32,[None,10])

	#add output layer
	# l1爲隱藏層，爲了更加看出overfitting，所以輸出給了100
	l1=add_layer_EX(xs,64,100,'l1',activation_function=tf.nn.tanh)

	prediction=add_layer_EX(l1,100,10,'l2',activation_function=tf.nn.softmax)

	#the error between prediction and real data
	cross_entropy=tf.reduce_mean(-tf.reduce_sum(ys*tf.log(prediction),reduction_indices=[1]))
	#添加標量統計結果
	tf.summary.scalar('loss',cross_entropy)
	train_step=tf.train.GradientDescentOptimizer(0.4).minimize(cross_entropy)

	init=tf.global_variables_initializer()

	with tf.Session() as sess:
		sess.run(init)
		#添加一個操作，代表執行所有summary操作，這樣可以避免人工執行每一個summary op
		merged=tf.summary.merge_all()
		#summary writer goes in here
		train_writer=tf.summary.FileWriter("logs/train",sess.graph)#train爲log的子文件夾
		test_writer=tf.summary.FileWriter("logs/test",sess.graph)
		for i in range(2500):
			#sess.run(train_step,feed_dict={xs:X_train,ys:y_train})
			# drop掉60%，保持40%不被drop掉
			sess.run(train_step,feed_dict={xs:X_train,ys:y_train,keep_prob:0.4})
			if i%50==0:
				#record loss
				y_pre=sess.run(prediction,feed_dict={xs:X_test})
				correct_prediction=tf.equal(tf.arg_max(y_pre,1),tf.arg_max(y_test,1))
				accuracy=tf.reduce_mean(tf.cast(correct_prediction,tf.float32))
				result=sess.run(accuracy,feed_dict={xs:X_test,ys:y_test})
				print("epoch ", i, ",accuracy:", result)
				
				train_result=sess.run(merged,feed_dict={xs:X_train,ys:y_train})
				test_result = sess.run(merged, feed_dict={xs: X_test, ys: y_test})
				train_writer.add_summary(train_result,i)
				test_writer.add_summary(test_result,i)

		
def main():
	#test01()
	#test02()
	#test03()
	#test04()
	#test05()
	#test06()
	#test07()
	#test08()
	#test09()
	#test10()
	test11()

if __name__ == '__main__':
	main()