Tensorflow擼程式碼之2邏輯迴歸
阿新 • • 發佈:2019-01-05
邏輯迴歸
詳細地址
# _*_ encoding=utf8 _*_
import tensorflow as tf
import numpy as np
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("data/", one_hot=True)
# 設定學習率,
learning_rate = 0.01
training_epochs = 25
batch_size = 100
display_step = 1
#圖片都是28*28 = 784的 預測為0-9 10個數字 10分類
x = tf.placeholder(tf.float32, [None, 784])
y = tf.placeholder(tf.float32, [None, 10])
# 初始化引數的值
W = tf.Variable(tf.zeros([28*28, 10]),name="W")
b = tf.Variable(tf.zeros([10]))
# 前向傳播
pred = tf.nn.softmax(tf.matmul(x, W) + b)
# 使用交叉熵作為損失函式
cost = tf.reduce_mean(-tf.reduce_sum(y*tf.log(pred), reduction_indices= 1))
# 梯度下降求最佳的cost
optimizer = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost)
init = tf.global_variables_initializer()
with tf.Session() as sess:
sess.run(init)
print(W) #<tf.Variable 'W:0' shape=(784, 10) dtype=float32_ref>
for epoch in range(training_epochs):
# 得到資料總的batch
total_batch = int(mnist.train.num_examples / batch_size)
# print(total_batch) #550 個batch 55000張訓練資料
# 每一訓練 都去求得平均的cost
avg_cost = 0
#batch訓練 遍歷所以得batch
for i in range(total_batch):
batch_x,batch_y = mnist.train.next_batch(batch_size)
# 訓練引數
_, c = sess.run([optimizer, cost], feed_dict={x: batch_x,y: batch_y})
avg_cost += c / total_batch
# 計算每一次迭代的cost
if (epoch + 1) % display_step == 0:
print("Epoch:", '%04d' % (epoch + 1), "cost=", "{:.9f}".format(avg_cost))
#評估模型 訓練出來的結果tf.argmax(pred, 1) 找10個數最大的那個 和真實值y對比
correct_prediction = tf.equal(tf.argmax(pred, 1), tf.argmax(y, 1))
# 計算準確率
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
# 使用mnist的test資料進行評估
# 首先執行之前的所有必要的操作來產生這個計算這個tensor需要的輸入,然後通過這些輸入產生這個tensor。
print("Accuracy:", accuracy.eval({x: mnist.test.images, y: mnist.test.labels}))