簡單python代碼實現三層神經網絡識別手寫數字

準備

這個過程非常簡單，就是用到了很多的矩陣運算。

• 訓練數據集下載地址，
• 測試數據集下載地址，

數據格式

.csv格式數據的每一行都是一個28*28像素的手寫數字圖片，每一行的第一個像素是數字的值，從第二個數字開始時像素值

 

import matplotlib.pyplot
import pylab
import numpy

# 讀入訓練數據
training_data_file = open("G:\神經網絡數據\mnist_train.csv", "r")
training_data_list = training_data_file.readlines()
training_data_file.close()
# 圖片展示
aist = training_data_list[1].split(",")
aist = numpy.asfarray(aist[1:]).reshape((28, 28))
matplotlib.pyplot.imshow(aist, interpolation="nearest")
pylab.show()

效果圖展示

神經網絡代碼

1. 對象

 

import numpy
import scipy.special


class NeuralNetWork:
    # 初始化
    def __init__(self, inputnodfes, hiddennodes, outputnodes, learningrate):
        self.innodes = inputnodfes
        self.hnodes = hiddennodes
        self.onodes = outputnodes
        self.lr = learningrate
        self.wih = numpy.random.normal(0.0, pow(self.innodes, -0.5), (self.hnodes, self.innodes))
        self.who = numpy.random.normal(0.0, pow(self.hnodes, -0.5), (self.onodes, self.hnodes))
        # 抑制函數
        self.activation_function = lambda x: scipy.special.expit(x)
        pass

     # 訓練
    def train(self, inputs_list, targets_list):
        inputs = numpy.array(inputs_list, ndmin=2).T
        targets = numpy.array(targets_list, ndmin=2).T
        hidden_inputs = numpy.dot(self.wih, inputs)
        hidden_outputs = self.activation_function(hidden_inputs)
        final_inputs = numpy.dot(self.who, hidden_outputs)
        final_outputs = self.activation_function(final_inputs)
        output_errors = targets - final_outputs
        hidden_errors = numpy.dot(self.who.T, output_errors)
        self.who += self.lr * numpy.dot((output_errors * final_outputs * (1.0 - final_outputs)),
                                        numpy.transpose(hidden_outputs))
        self.wih += self.lr * numpy.dot((hidden_errors * hidden_outputs * (1.0 - hidden_outputs)),
                                        numpy.transpose(inputs))
        pass

    # 測試
    def query(self, inputs_list):
        inputs = numpy.array(inputs_list, ndmin=2).T
        hidden_inputs = numpy.dot(self.wih, inputs)
        hidden_outputs = self.activation_function(hidden_inputs)
        final_inputs = numpy.dot(self.who, hidden_outputs)
        final_outputs = self.activation_function(final_inputs)
        return final_outputs

1. 訓練和測試代碼

 

import numpy

from neuralNetwork import NeuralNetWork


input_nodes = 784
hidden_nodes = 100
output_nodes = 10
learning_rate = 0.2

# 讀入訓練數據
training_data_file = open("G:\神經網絡數據\mnist_train.csv", "r")
training_data_list = training_data_file.readlines()
training_data_file.close()
# 初始化神經網絡
b = NeuralNetWork(input_nodes, hidden_nodes, output_nodes, learning_rate)

for record in training_data_list:
    all_values = record.split(',')
    inputs = (numpy.asfarray(all_values[1:]) / 255.0 * 0.9) + 0.01
    targets = numpy.zeros(output_nodes) + 0.01
    targets[int(all_values[0])] = 0.99
    b.train(inputs, targets)
    pass

print("訓練完成")
# 使用測試數據測試準確性
test_data_file = open("G:\神經網絡數據\mnist_test.csv", "r")
test_data_list = test_data_file.readlines()
test_data_file.close()
# 使用算法逐項對比測試數據是否準確，然後統計
scorecard = []
for record in test_data_list:
    all_values = record.split(',')
    correct_label = int(all_values[0])
    inputs = (numpy.asfarray(all_values[1:]) / 255.0 * 0.99) + 0.01
    outputs = b.query(inputs)
    label = numpy.argmax(outputs)
    if label == correct_label:
        scorecard.append(1)
    else:
        scorecard.append(0)
        pass
    pass

scorecard_array = numpy.asarray(scorecard)
print("準確率", scorecard_array.sum() / scorecard_array.size)

結果

最後運行的結果顯示準確率達到94%