TensorFlow 核心概念:
計算表示爲一個有向圖(directed graph),或計算圖(computation graph)
其中每一步運算操作(operation)作爲一個節點(node)
節點與節點之間的連線成爲邊(edge)
在計算圖邊中流動(flow)的數據成爲張量(tensor)
TensorFlow神經網絡開發流程:
1、定義算法公式,也就是神經網絡forward是的計算
2、定義loss,選定優化器,並指定優化器優化loss
3、迭代地對數據進行訓練
4、在測試集或驗證集上對準確率進行評測
運用softmax 的BP網絡訓練,代碼如下:
#下載引入數據集
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)
import tensorflow as tf
#設置默認會話(session)
sess = tf.InteractiveSession()
#設置模型變量
x = tf.placeholder(tf.float32, [None, 784])
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))
y = tf.nn.softmax(tf.matmul(x, W) + b)
#訓練模型
#成本函數:交叉熵
y_ = tf.placeholder(tf.float32, [None, 10])
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices = [1]))
train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy)
#初始化變量
init = tf.initialize_all_variables()
#啓動圖(graph)
sess = tf.Session()
sess.run(init)
#訓練模型1000次
for i in range(40000):
batch_xs, batch_ys = mnist.train.next_batch(100)
sess.run(train_step, feed_dict={x:batch_xs,y_:batch_ys})
print(sess.run(W),sess.run(b))
#評估模型
correct_prediction = tf.equal(tf.argmax(y,1), tf.argmax(y_,1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction,"float"))
print(sess.run(accuracy,feed_dict={x:mnist.test.images, y_:mnist.test.labels}))