秦凱新技術社群推出的《Coding技術進階實戰》系列即將上線，包含語言類精深的用法和技巧，涵蓋 python,Java,Scala,Tensorflow等主流大資料和深度學習技術基礎，敬請期待。為什麼我會寫這樣一個系列，來源於被一位容器雲專家問到如何實現一個執行緒池時，讓我頓感以前研究的Java併發控制相關的理論以及多執行緒併發設計模式忘得九霄雲外，鑑於此，氣憤難平，決定展示個人程式設計魅力。
版權宣告：本套技術專欄是作者（秦凱新）平時工作的總結和昇華，通過從真實商業環境抽取案例進行總結和分享，並給出商業應用的調優建議和叢集環境容量規劃等內容，請持續關注本套部落格。QQ郵箱地址：1120746959@qq.com，如有任何技術交流，可隨時聯絡。
1 TensorFlow基本使用操作
-
TensorFlow基本模型
import tensorflow as tf a = 3 # Create a variable. w = tf.Variable([[0.5,1.0]]) x = tf.Variable([[2.0],[1.0]]) y = tf.matmul(w, x) #variables have to be explicitly initialized before you can run Ops init_op = tf.global_variables_initializer() with tf.Session() as sess: sess.run(init_op) print (y.eval()) 複製程式碼 -
TensorFlow基本資料型別
# float32 tf.zeros([3, 4], int32) ==> [[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0]] # 'tensor' is [[1, 2, 3], [4, 5, 6]] tf.zeros_like(tensor) ==> [[0, 0, 0], [0, 0, 0]] tf.ones([2, 3], int32) ==> [[1, 1, 1], [1, 1, 1]] # 'tensor' is [[1, 2, 3], [4, 5, 6]] tf.ones_like(tensor) ==> [[1, 1, 1], [1, 1, 1]] # Constant 1-D Tensor populated with value list. tensor = tf.constant([1, 2, 3, 4, 5, 6, 7]) => [1 2 3 4 5 6 7] # Constant 2-D tensor populated with scalar value -1. tensor = tf.constant(-1.0, shape=[2, 3]) => [[-1. -1. -1.] [-1. -1. -1.]] tf.linspace(10.0, 12.0, 3, name="linspace") => [ 10.0 11.0 12.0] # 'start' is 3 # 'limit' is 18 # 'delta' is 3 tf.range(start, limit, delta) ==> [3, 6, 9, 12, 15] 複製程式碼 -
random_shuffle運算元及random_normal運算元
norm = tf.random_normal([2, 3], mean=-1, stddev=4) # Shuffle the first dimension of a tensor c = tf.constant([[1, 2], [3, 4], [5, 6]]) shuff = tf.random_shuffle(c) # Each time we run these ops, different results are generated sess = tf.Session() print (sess.run(norm)) print (sess.run(shuff)) [[-0.30886292 3.11809683 3.29861784] [-7.09597015 -1.89811802 1.75282788]] [[3 4] [5 6] [1 2]] 複製程式碼 -
簡單操作的複雜性
state = tf.Variable(0) new_value = tf.add(state, tf.constant(1)) update = tf.assign(state, new_value) with tf.Session() as sess: sess.run(tf.global_variables_initializer()) print(sess.run(state)) for _ in range(3): sess.run(update) print(sess.run(state)) 複製程式碼 -
模型的儲存與載入
#tf.train.Saver w = tf.Variable([[0.5,1.0]]) x = tf.Variable([[2.0],[1.0]]) y = tf.matmul(w, x) init_op = tf.global_variables_initializer() saver = tf.train.Saver() with tf.Session() as sess: sess.run(init_op) # Do some work with the model. # Save the variables to disk. save_path = saver.save(sess, "C://tensorflow//model//test") print ("Model saved in file: ", save_path) 複製程式碼 -
numpy與TensorFlow互轉
import numpy as np a = np.zeros((3,3)) ta = tf.convert_to_tensor(a) with tf.Session() as sess: print(sess.run(ta)) 複製程式碼 -
TensorFlow佔坑操作
input1 = tf.placeholder(tf.float32) input2 = tf.placeholder(tf.float32) output = tf.mul(input1, input2) with tf.Session() as sess: print(sess.run([output], feed_dict={input1:[7.], input2:[2.]})) 複製程式碼
2 TensorFlow線性迴歸實現
-
numpy線性資料集生成
import numpy as np import tensorflow as tf import matplotlib.pyplot as plt # 隨機生成1000個點，圍繞在y=0.1x+0.3的直線周圍 num_points = 1000 vectors_set = [] for i in range(num_points): x1 = np.random.normal(0.0, 0.55) y1 = x1 * 0.1 + 0.3 + np.random.normal(0.0, 0.03) vectors_set.append([x1, y1]) # 生成一些樣本 x_data = [v[0] for v in vectors_set] y_data = [v[1] for v in vectors_set] plt.scatter(x_data,y_data,c='r') plt.show() 複製程式碼
-
TensorFlow實現線性模型
生成1維的W矩陣，取值是[-1,1]之間的隨機數 W = tf.Variable(tf.random_uniform([1], -1.0, 1.0), name='W') # 生成1維的b矩陣，初始值是0 b = tf.Variable(tf.zeros([1]), name='b') # 經過計算得出預估值y y = W * x_data + b # Loss: 以預估值y和實際值y_data之間的均方誤差作為損失 loss = tf.reduce_mean(tf.square(y - y_data), name='loss') # 優化器：採用梯度下降法來優化引數（train模組，參數列示學習率） optimizer = tf.train.GradientDescentOptimizer(0.5) # 開始訓練：訓練的過程就是最小化這個誤差值 train = optimizer.minimize(loss, name='train') sess = tf.Session() init = tf.global_variables_initializer() sess.run(init) # 初始化的W和b是多少 print ("W =", sess.run(W), "b =", sess.run(b), "loss =", sess.run(loss)) # 執行20次訓練 for step in range(20): sess.run(train) # 輸出訓練好的W和b print ("W =", sess.run(W), "b =", sess.run(b), "loss =", sess.run(loss)) writer = tf.train.SummaryWriter("./tmp", sess.graph) 複製程式碼 -
TensorFlow迭代結果
W = [ 0.96539688] b = [ 0.] loss = 0.297884 W = [ 0.71998411] b = [ 0.28193575] loss = 0.112606 W = [ 0.54009342] b = [ 0.28695393] loss = 0.0572231 W = [ 0.41235447] b = [ 0.29063231] loss = 0.0292957 W = [ 0.32164571] b = [ 0.2932443] loss = 0.0152131 W = [ 0.25723246] b = [ 0.29509908] loss = 0.00811188 W = [ 0.21149193] b = [ 0.29641619] loss = 0.00453103 W = [ 0.17901111] b = [ 0.29735151] loss = 0.00272536 W = [ 0.15594614] b = [ 0.29801565] loss = 0.00181483 W = [ 0.13956745] b = [ 0.29848731] loss = 0.0013557 W = [ 0.12793678] b = [ 0.29882219] loss = 0.00112418 W = [ 0.11967772] b = [ 0.29906002] loss = 0.00100743 W = [ 0.11381286] b = [ 0.29922891] loss = 0.000948558 W = [ 0.10964818] b = [ 0.29934883] loss = 0.000918872 W = [ 0.10669079] b = [ 0.29943398] loss = 0.000903903 W = [ 0.10459071] b = [ 0.29949448] loss = 0.000896354 W = [ 0.10309943] b = [ 0.29953739] loss = 0.000892548 W = [ 0.10204045] b = [ 0.29956791] loss = 0.000890629 W = [ 0.10128847] b = [ 0.29958954] loss = 0.000889661 W = [ 0.10075447] b = [ 0.29960492] loss = 0.000889173 W = [ 0.10037527] b = [ 0.29961586] loss = 0.000888927 plt.scatter(x_data,y_data,c='r') plt.plot(x_data,sess.run(W)*x_data+sess.run(b)) plt.show() 複製程式碼 -
版權宣告：本套技術專欄是作者（秦凱新）平時工作的總結和昇華，通過從真實商業環境抽取案例進行總結和分享，並給出商業應用的調優建議和叢集環境容量規劃等內容，請持續關注本套部落格。QQ郵箱地址：1120746959@qq.com，如有任何學術交流，可隨時聯絡。
3 MNIST資料集載入介紹
-
載入
import numpy as np import tensorflow as tf import matplotlib.pyplot as plt #from tensorflow.examples.tutorials.mnist import input_data import input_data print ("packs loaded") print ("Download and Extract MNIST dataset") ##使用one_hot 01編碼 mnist = input_data.read_data_sets('data/', one_hot=True) print print (" tpye of 'mnist' is %s" % (type(mnist))) print (" number of trian data is %d" % (mnist.train.num_examples)) print (" number of test data is %d" % (mnist.test.num_examples)) Download and Extract MNIST dataset Extracting data/train-images-idx3-ubyte.gz Extracting data/train-labels-idx1-ubyte.gz Extracting data/t10k-images-idx3-ubyte.gz Extracting data/t10k-labels-idx1-ubyte.gz tpye of 'mnist' is <class 'tensorflow.contrib.learn.python.learn.datasets.base.Datasets'> number of trian data is 55000 number of test data is 10000 複製程式碼 -
What does the data of MNIST look like?
print ("What does the data of MNIST look like?") trainimg = mnist.train.images trainlabel = mnist.train.labels testimg = mnist.test.images testlabel = mnist.test.labels print print (" type of 'trainimg' is %s" % (type(trainimg))) print (" type of 'trainlabel' is %s" % (type(trainlabel))) print (" type of 'testimg' is %s" % (type(testimg))) print (" type of 'testlabel' is %s" % (type(testlabel))) print (" shape of 'trainimg' is %s" % (trainimg.shape,)) print (" shape of 'trainlabel' is %s" % (trainlabel.shape,)) print (" shape of 'testimg' is %s" % (testimg.shape,)) print (" shape of 'testlabel' is %s" % (testlabel.shape,)) What does the data of MNIST look like? type of 'trainimg' is <class 'numpy.ndarray'> type of 'trainlabel' is <class 'numpy.ndarray'> type of 'testimg' is <class 'numpy.ndarray'> type of 'testlabel' is <class 'numpy.ndarray'> shape of 'trainimg' is (55000, 784) shape of 'trainlabel' is (55000, 10) shape of 'testimg' is (10000, 784) shape of 'testlabel' is (10000, 10) 複製程式碼 -
How does the training data look like?
# How does the training data look like? print ("How does the training data look like?") nsample = 5 randidx = np.random.randint(trainimg.shape[0], size=nsample) for i in randidx: curr_img = np.reshape(trainimg[i, :], (28, 28)) # 28 by 28 matrix curr_label = np.argmax(trainlabel[i, :] ) # Label plt.matshow(curr_img, cmap=plt.get_cmap('gray')) plt.title("" + str(i) + "th Training Data " + "Label is " + str(curr_label)) print ("" + str(i) + "th Training Data " + "Label is " + str(curr_label)) plt.show() 複製程式碼
-
Batch Learning?
print ("Batch Learning? ") batch_size = 100 batch_xs, batch_ys = mnist.train.next_batch(batch_size) print ("type of 'batch_xs' is %s" % (type(batch_xs))) print ("type of 'batch_ys' is %s" % (type(batch_ys))) print ("shape of 'batch_xs' is %s" % (batch_xs.shape,)) print ("shape of 'batch_ys' is %s" % (batch_ys.shape,)) Batch Learning? type of 'batch_xs' is <class 'numpy.ndarray'> type of 'batch_ys' is <class 'numpy.ndarray'> shape of 'batch_xs' is (100, 784) shape of 'batch_ys' is (100, 10) 複製程式碼
4 MNIST資料集邏輯迴歸測試
-
tensorflow的tf.reduce_mean函式
m1 = tf.reduce_mean(x, axis=0) 結果為：[1.5, 1.5] 複製程式碼 -
tensorflow的argmaxtensorflow的 sess = tf.InteractiveSession()
arr = np.array([[31, 23, 4, 24, 27, 34], [18, 3, 25, 0, 6, 35], [28, 14, 33, 22, 20, 8], [13, 30, 21, 19, 7, 9], [16, 1, 26, 32, 2, 29], [17, 12, 5, 11, 10, 15]]) #列印加上eval ## 矩陣的維度 2 #tf.rank(arr).eval() ## 矩陣行和列 [6,6] #tf.shape(arr).eval() # 引數0表示維度，按照列。 表示最每列最大值的索引 [0,3,2,4,0,1] #tf.argmax(arr, 0).eval() # 0 -> 31 (arr[0, 0]) # 3 -> 30 (arr[3, 1]) # 2 -> 33 (arr[2, 2]) tf.argmax(arr, 1).eval() # 5 -> 34 (arr[0, 5]) # 5 -> 35 (arr[1, 5]) # 2 -> 33 (arr[2, 2]) array([5, 5, 2, 1, 3, 0], dtype=int64) 複製程式碼 -
載入資料集
import numpy as np import tensorflow as tf import matplotlib.pyplot as plt import input_data mnist = input_data.read_data_sets('data/', one_hot=True) trainimg = mnist.train.images trainlabel = mnist.train.labels testimg = mnist.test.images testlabel = mnist.test.labels print ("MNIST loaded") Extracting data/train-images-idx3-ubyte.gz Extracting data/train-labels-idx1-ubyte.gz Extracting data/t10k-images-idx3-ubyte.gz Extracting data/t10k-labels-idx1-ubyte.gz MNIST loaded print (trainimg.shape) print (trainlabel.shape) print (testimg.shape) print (testlabel.shape) #print (trainimg) print (trainlabel[0]) (55000, 784) (55000, 10) (10000, 784) (10000, 10) [ 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.] 複製程式碼 -
TF邏輯迴歸模型構建
# 先放坑（每一行是一個樣本） x = tf.placeholder("float", [None, 784]) # 總共10位 [ 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.] y = tf.placeholder("float", [None, 10]) # None is for infinite #10分類任務 784輸入，10代表輸出 W = tf.Variable(tf.zeros([784, 10])) # 10代表輸出 b = tf.Variable(tf.zeros([10])) # LOGISTIC REGRESSION MODEL（輸出為10） actv = tf.nn.softmax(tf.matmul(x, W) + b) # COST FUNCTION（損失函式） cost = tf.reduce_mean(-tf.reduce_sum(y*tf.log(actv), reduction_indices=1)) # OPTIMIZER learning_rate = 0.01 optm = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost) 複製程式碼 -
TF模型訓練
##迭代次數 training_epochs = 50 每次迭代多少樣本 batch_size = 100 display_step = 5 # SESSION sess = tf.Session() sess.run(init) # MINI-BATCH LEARNING for epoch in range(training_epochs): avg_cost = 0. num_batch = int(mnist.train.num_examples/batch_size) for i in range(num_batch): batch_xs, batch_ys = mnist.train.next_batch(batch_size) sess.run(optm, feed_dict={x: batch_xs, y: batch_ys}) feeds = {x: batch_xs, y: batch_ys} avg_cost += sess.run(cost, feed_dict=feeds)/num_batch # DISPLAY if epoch % display_step == 0: feeds_train = {x: batch_xs, y: batch_ys} feeds_test = {x: mnist.test.images, y: mnist.test.labels} train_acc = sess.run(accr, feed_dict=feeds_train) test_acc = sess.run(accr, feed_dict=feeds_test) print ("Epoch: %03d/%03d cost: %.9f train_acc: %.3f test_acc: %.3f" % (epoch, training_epochs, avg_cost, train_acc, test_acc)) print ("DONE") Epoch: 000/050 cost: 1.177906594 train_acc: 0.840 test_acc: 0.855 Epoch: 005/050 cost: 0.440515266 train_acc: 0.860 test_acc: 0.895 Epoch: 010/050 cost: 0.382895913 train_acc: 0.910 test_acc: 0.905 Epoch: 015/050 cost: 0.356607343 train_acc: 0.870 test_acc: 0.909 Epoch: 020/050 cost: 0.341326642 train_acc: 0.860 test_acc: 0.912 Epoch: 025/050 cost: 0.330556413 train_acc: 0.910 test_acc: 0.913 Epoch: 030/050 cost: 0.321508561 train_acc: 0.840 test_acc: 0.916 Epoch: 035/050 cost: 0.314936944 train_acc: 0.940 test_acc: 0.917 Epoch: 040/050 cost: 0.309805418 train_acc: 0.940 test_acc: 0.918 Epoch: 045/050 cost: 0.305343132 train_acc: 0.960 test_acc: 0.918 DONE 複製程式碼
5 總結
通過簡單的案例，真正明白TensorFlow設計思想，才是本文的目的。
版權宣告：本套技術專欄是作者（秦凱新）平時工作的總結和昇華，通過從真實商業環境抽取案例進行總結和分享，並給出商業應用的調優建議和叢集環境容量規劃等內容，請持續關注本套部落格。QQ郵箱地址：1120746959@qq.com，如有任何學術交流，可隨時聯絡。
秦凱新 於深圳 201812092128