笔记汇总：https://blog.csdn.net/a1103688841/article/details/84350432
网络搭建参考：https://blog.csdn.net/a1103688841/article/details/85145456
V1 版本----使用100训练集和10测试集进行初步的代码验证
# -*- coding: UTF-8 -*-
import numpy as np
import scipy.special
import matplotlib.pyplot
#不以科学计数法显示数据
np.set_printoptions(suppress=True)
class neuralNetwork:
def __init__(self,inputnodes,hiddennodes,outputnodes,learningrate,weit_simple=True):
#basis value initialize
self.inodes=inputnodes
self.hnodes=hiddennodes
self.onodes=outputnodes
self.lr=learningrate
#weit value initialize
if(weit_simple):
self.wih=(np.random.rand(self.hnodes,self.inodes)-0.5)
self.who=(np.random.rand(self.onodes,self.hnodes)-0.5)
else:
self.wih=np.random.normal(0.0,pow(self.hnodes,-0.5),(self.hnodes,self.inodes))
self.who=np.random.normal(0.0, pow(self.onodes, -0.5), (self.onodes, self.hnodes))
#activation function initialize
#这里不能随意修改函数，因为下面训练函数中的激活函数导数是直接固定的。
self.activaltion_function=lambda x:scipy.special.expit(x)
pass
def train(self,inputs_list,targets_list):
inputs=np.array(inputs_list,ndmin=2).T
targets=np.array(targets_list,ndmin=2).T
hidden_inputs=np.dot(self.wih,inputs)
hidden_outputs=self.activaltion_function(hidden_inputs)
final_inputs=np.dot(self.who,hidden_outputs)
final_outputs=self.activaltion_function(final_inputs)
output_errors=targets-final_outputs
hidden_errors=np.dot(self.who.T,output_errors)
#这里使用的反向传播用的是S型函数的公式
self.who+=self.lr*np.dot((output_errors*final_outputs*(1.0-final_outputs)),np.transpose(hidden_outputs))
self.wih += self.lr * np.dot((hidden_errors * hidden_outputs * (1.0 - hidden_outputs)),np.transpose(inputs))
pass
def query(self,inputs_list):
inputs=np.array(inputs_list,ndmin=2).T
hidden_inputs=np.dot(self.wih,inputs)
hidden_outputs=self.activaltion_function(hidden_inputs)
final_inputs=np.dot(self.who,hidden_outputs)
final_outputs=self.activaltion_function(final_inputs)
return final_outputs
input_nodes = 784
hidden_nodes = 200
output_nodes = 10
learning_rate = 0.3
n = neuralNetwork(input_nodes,hidden_nodes,output_nodes, learning_rate)
#读取数据
training_data_file = open("mnist_train_100.csv", 'r')
training_data_list = training_data_file.readlines()
training_data_file.close()
for record in training_data_list:
#数据初始化
all_values = record.split(',')
inputs = (np.asfarray(all_values[1:]) / 255.0 * 0.99) + 0.01
#转独热码
targets = np.zeros(output_nodes) + 0.01
targets[int(all_values[0])] = 0.99
n.train(inputs, targets)
pass
#测试数据集读取
test_data_file = open("mnist_test_10.csv", 'r')
test_data_list = test_data_file.readlines()
test_data_file.close()
all_values=test_data_list[0].split(',')
#目前测试输入的数值
print(all_values[0])
image_array=np.asfarray(all_values[1:]).reshape((28,28))
matplotlib.pyplot.imshow(image_array,cmap='Greys',interpolation='None')
#matplotlib.pyplot.show()
output=n.query((np.asfarray(all_values[1:])/255.0*0.99)+0.01)
#最后看独热编码的哪一位数字大就是输出几
print(output)V2 版本----加入记分牌和5个世代的训练
# -*- coding: UTF-8 -*-
import numpy as np
import scipy.special
import matplotlib.pyplot
#不以科学计数法显示数据
np.set_printoptions(suppress=True)
class neuralNetwork:
def __init__(self,inputnodes,hiddennodes,outputnodes,learningrate,weit_simple=True):
#basis value initialize
self.inodes=inputnodes
self.hnodes=hiddennodes
self.onodes=outputnodes
self.lr=learningrate
#weit value initialize
if(weit_simple):
self.wih=(np.random.rand(self.hnodes,self.inodes)-0.5)
self.who=(np.random.rand(self.onodes,self.hnodes)-0.5)
else:
self.wih=np.random.normal(0.0,pow(self.hnodes,-0.5),(self.hnodes,self.inodes))
self.who=np.random.normal(0.0, pow(self.onodes, -0.5), (self.onodes, self.hnodes))
#activation function initialize
#这里不能随意修改函数，因为下面训练函数中的激活函数导数是直接固定的。
self.activaltion_function=lambda x:scipy.special.expit(x)
pass
def train(self,inputs_list,targets_list):
inputs=np.array(inputs_list,ndmin=2).T
targets=np.array(targets_list,ndmin=2).T
hidden_inputs=np.dot(self.wih,inputs)
hidden_outputs=self.activaltion_function(hidden_inputs)
final_inputs=np.dot(self.who,hidden_outputs)
final_outputs=self.activaltion_function(final_inputs)
output_errors=targets-final_outputs
hidden_errors=np.dot(self.who.T,output_errors)
#这里使用的反向传播用的是S型函数的公式
self.who+=self.lr*np.dot((output_errors*final_outputs*(1.0-final_outputs)),np.transpose(hidden_outputs))
self.wih += self.lr * np.dot((hidden_errors * hidden_outputs * (1.0 - hidden_outputs)),np.transpose(inputs))
pass
def query(self,inputs_list):
inputs=np.array(inputs_list,ndmin=2).T
hidden_inputs=np.dot(self.wih,inputs)
hidden_outputs=self.activaltion_function(hidden_inputs)
final_inputs=np.dot(self.who,hidden_outputs)
final_outputs=self.activaltion_function(final_inputs)
return final_outputs
input_nodes = 784
hidden_nodes = 200
output_nodes = 10
learning_rate = 0.2
n = neuralNetwork(input_nodes,hidden_nodes,output_nodes, learning_rate)
#读取数据
training_data_file = open("mnist_train_100.csv", 'r')
training_data_list = training_data_file.readlines()
training_data_file.close()
#进行5世代训练
epochs = 5
for e in range(epochs):
for record in training_data_list:
#数据初始化
all_values = record.split(',')
inputs = (np.asfarray(all_values[1:]) / 255.0 * 0.99) + 0.01
#转独热码
targets = np.zeros(output_nodes) + 0.01
targets[int(all_values[0])] = 0.99
n.train(inputs, targets)
pass
pass
test_data_file = open("mnist_test_10.csv", 'r')
test_data_list = test_data_file.readlines()
test_data_file.close()
#记分牌
scorecard=[]
for record in test_data_list:
all_values = record.split(',')
correct_label = int(all_values[0])
print(correct_label,"correct label")
inputs = (np.asfarray(all_values[1:]) / 255.0 * 0.99) + 0.01
outputs = n.query(inputs)
label = np.argmax(outputs)
print(label,"network's answer")
if (label == correct_label):
scorecard.append(1)
else:
scorecard.append(0)
pass
pass
print(scorecard)
#计算准确率
scorecard_array = np.asarray(scorecard)
print ("performance = ", scorecard_array.sum()/float(scorecard_array.size))结果展示：
对于MNIST手写体超参数的优化
1）学习率部分
2）世代数目部分
3）世代数目和学习率共同作用部分
4）隐藏层节点数部分
