《神经网络与深度学习》作业github
来复习一遍
full connection numpy
# -*- coding: utf-8 -*-
import numpy as np
from tensorflow.examples.tutorials.mnist import input_data
class Model():
def __init__(self):
print("init...")
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
self.n_in = 784 # 每张图片是 28 *28 像素
self.n_out = 10 # 总共有 10个类
self.max_epochs = 10000 # 最大训练步数 1000步
self.Weights = np.random.rand(self.n_in,self.n_out) # initialize W 0
self.biases = np.zeros(self.n_out) # initialize bias 0
for i in range(self.max_epochs):
batch_xs, batch_ys = mnist.train.next_batch(100)
batch_xs = np.array(batch_xs)
batch_ys = np.array(batch_ys)
self.train(batch_xs, batch_ys, 0.0001)
if i % 500 == 0:
accuracy_test = self.compute_accuracy(np.array(mnist.test.images[:500]), np.array(mnist.test.labels[:500]))
print("#"*30)
print("compute_accuracy:",accuracy_test)
print("cross_entropy:",self.cross_entropy(batch_ys,self.output(batch_xs) )) # 输出交叉熵损失函数
def train(self,batch_x,batch_y, learning_rate):# 训练数据 更新权重
#在下面补全（注意对齐空格）
y_pred = self.output(batch_x)
grad = np.dot(batch_x.T,(y_pred-batch_y))
# print(batch_x.shape)
# print(batch_y.shape)
# print(grad.shape)
# print(self.Weights.shape)
self.Weights = self.Weights - grad*learning_rate
def output(self, batch_x):# 输出预测值
# 注意防止 上溢出和下溢出
def softmax(x):
e_x = np.exp(x-np.max(x))
return e_x / (e_x.sum(axis=0)) +1e-30 #
prediction = np.add(np.dot(batch_x, self.Weights),self.biases)
result =[]
for i in range(len(prediction)):
result.append(softmax(prediction[i]))
return np.array(result)
def cross_entropy(self,batch_y,prediction_y): #交叉熵函数
cross_entropy = - np.mean(
np.sum(batch_y * np.log(prediction_y),axis=1))
return cross_entropy
def compute_accuracy(self,xs, ys):# 计算预测精度
pre_y = self.output(xs)
pre_y_index = np.argmax(pre_y,axis =1)
y_index = np.argmax(ys,axis =1)
count_equal = np.equal(y_index,pre_y_index)
count = np.sum([1 for e in count_equal if e ])
sum_count =len(xs)
return count * 1.0 / sum_count
m = Model()
init...
WARNING:tensorflow:From :4: read_data_sets (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.
Instructions for updating:
Please use alternatives such as official/mnist/dataset.py from tensorflow/models.
WARNING:tensorflow:From /Users/gaohanning/.pyenv/versions/3.6.6/envs/dl/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:260: maybe_download (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.
Instructions for updating:
Please write your own downloading logic.
WARNING:tensorflow:From /Users/gaohanning/.pyenv/versions/3.6.6/envs/dl/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:262: extract_images (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.
Instructions for updating:
Please use tf.data to implement this functionality.
Extracting MNIST_data/train-images-idx3-ubyte.gz
WARNING:tensorflow:From /Users/gaohanning/.pyenv/versions/3.6.6/envs/dl/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:267: extract_labels (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.
Instructions for updating:
Please use tf.data to implement this functionality.
Extracting MNIST_data/train-labels-idx1-ubyte.gz
WARNING:tensorflow:From /Users/gaohanning/.pyenv/versions/3.6.6/envs/dl/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:110: dense_to_one_hot (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.
Instructions for updating:
Please use tf.one_hot on tensors.
Extracting MNIST_data/t10k-images-idx3-ubyte.gz
Extracting MNIST_data/t10k-labels-idx1-ubyte.gz
WARNING:tensorflow:From /Users/gaohanning/.pyenv/versions/3.6.6/envs/dl/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:290: DataSet.__init__ (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.
Instructions for updating:
Please use alternatives such as official/mnist/dataset.py from tensorflow/models.
##############################
compute_accuracy: 0.084
cross_entropy: 5.217768517589507
##############################
compute_accuracy: 0.496
cross_entropy: 1.3667549956884744
##############################
compute_accuracy: 0.696
cross_entropy: 0.9100604725017135
##############################
compute_accuracy: 0.756
cross_entropy: 0.5283512200394022
##############################
compute_accuracy: 0.786
cross_entropy: 0.5764147291756632
##############################
compute_accuracy: 0.804
cross_entropy: 0.37392200139605103
##############################
compute_accuracy: 0.832
cross_entropy: 0.6398413262851023
##############################
compute_accuracy: 0.834
cross_entropy: 0.5282240686661552
##############################
compute_accuracy: 0.846
cross_entropy: 0.37446538466513324
##############################
compute_accuracy: 0.848
cross_entropy: 0.7063916594615859
##############################
compute_accuracy: 0.854
cross_entropy: 0.4532571558888631
##############################
compute_accuracy: 0.866
cross_entropy: 0.5824548984274599
##############################
compute_accuracy: 0.864
cross_entropy: 0.4608331339898249
##############################
compute_accuracy: 0.866
cross_entropy: 0.3215719108943983
##############################
compute_accuracy: 0.866
cross_entropy: 0.47378736687496287
##############################
compute_accuracy: 0.87
cross_entropy: 0.407303310977041
##############################
compute_accuracy: 0.872
cross_entropy: 0.342605986543631
##############################
compute_accuracy: 0.874
cross_entropy: 0.29502702721525553
##############################
compute_accuracy: 0.874
cross_entropy: 0.5693335527019046
##############################
compute_accuracy: 0.876
cross_entropy: 0.41044814982575467
full connection tensorflow
输入：
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
# number 1 to 10 data
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
输出：
WARNING:tensorflow:From :4: read_data_sets (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.
Instructions for updating:
Please use alternatives such as official/mnist/dataset.py from tensorflow/models.
WARNING:tensorflow:From /Users/gaohanning/.pyenv/versions/3.6.6/envs/dl/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:260: maybe_download (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.
Instructions for updating:
Please write your own downloading logic.
WARNING:tensorflow:From /Users/gaohanning/.pyenv/versions/3.6.6/envs/dl/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:262: extract_images (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.
Instructions for updating:
Please use tf.data to implement this functionality.
Extracting MNIST_data/train-images-idx3-ubyte.gz
WARNING:tensorflow:From /Users/gaohanning/.pyenv/versions/3.6.6/envs/dl/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:267: extract_labels (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.
Instructions for updating:
Please use tf.data to implement this functionality.
Extracting MNIST_data/train-labels-idx1-ubyte.gz
WARNING:tensorflow:From /Users/gaohanning/.pyenv/versions/3.6.6/envs/dl/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:110: dense_to_one_hot (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.
Instructions for updating:
Please use tf.one_hot on tensors.
Extracting MNIST_data/t10k-images-idx3-ubyte.gz
Extracting MNIST_data/t10k-labels-idx1-ubyte.gz
WARNING:tensorflow:From /Users/gaohanning/.pyenv/versions/3.6.6/envs/dl/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:290: DataSet.__init__ (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.
Instructions for updating:
Please use alternatives such as official/mnist/dataset.py from tensorflow/models.
输入：
def compute_accuracy(v_xs, v_ys):
global prediction
y_pre = sess.run(prediction, feed_dict={inputs: v_xs})
correct_prediction = tf.equal(tf.argmax(y_pre,1), tf.argmax(v_ys,1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
result = sess.run(accuracy, feed_dict={inputs: v_xs, outputs: v_ys})
return result
#定义相关参数
in_size = 784 # 输入的尺寸
out_size = 10 # 输出的尺寸
learning_rate = 0.1 #学习率
max_epochs = 5000 # 最大训练步数
inputs = tf.placeholder(tf.float32, [None, in_size]) # 28x28
outputs = tf.placeholder(tf.float32, [None, out_size])# 10
Weights = tf.Variable(tf.random_normal([in_size, out_size]))
biases = tf.Variable(tf.zeros([1, out_size]) + 0.1,)
prediction = tf.nn.softmax(tf.matmul(inputs, Weights) + biases)
#################################################################
# 请在下面补全交叉熵 和 训练两个节点。
cross_entropy = tf.nn.softmax_cross_entropy_with_logits(labels=outputs,logits=prediction)
train_step = tf.train.GradientDescentOptimizer(learning_rate=learning_rate).minimize(cross_entropy)
#################################################################
sess = tf.Session()
init = tf.global_variables_initializer() # 注意tensorflow 版本在 0.12 以后，才能执行这句命令
sess.run(init)
for i in range(max_epochs):
batch_xs, batch_ys = mnist.train.next_batch(100)
sess.run(train_step, feed_dict={inputs: batch_xs, outputs: batch_ys})
if i % 500 == 0:
print(compute_accuracy(
mnist.test.images, mnist.test.labels))
输出：
0.15
0.7789
0.8174
0.8241
0.8288
0.8327
0.8326
0.8336
0.8285
0.8362