keras快速搭建神经网络实现手写数字识别

利用keras 快速搭建神经网络模型，实现手写数字识别。用tensorflow作为后端

代码

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from scipy.io import loadmat
from sklearn.model_selection import train_test_split
import keras
from keras.models import Sequential
from keras.layers import Dense

def load_data(path):
    data = loadmat(path)
    X = data['X']
    y = data['y']
    return X,y

X, y = load_data('C:\\Users\\21153\Desktop\\ML作业\\ex3-neural network\\ex3data1.mat')
print(np.unique(y))  # 看下有几类标签
#y = y.flatten()       # 对y进行降维
X.shape, y.shape

def realdata(y,k):
    real = np.zeros(shape=(k,len(y)))
    for i in range(1,k+1):
        y_i = np.array([1 if label == i else 0 for label in y])
        real[i-1] = y_i
    return real.T
y = realdata(y,10)

X_train,X_test,y_train,y_test = train_test_split(X,y,test_size=0.3)
#y_train=y_train.T
X_train.shape,y_train.shape

# 搭建神经网络
model = Sequential()
model.add(Dense(units=16, activation='relu',input_dim = 400))   # 输入层，16个激活单元，激活函数为relu，输入数据维度为（400，）
model.add(Dense(units=16, activation='relu'))                 # 隐藏层，16个激活单元，激活函数为relu
model.add(Dense(units=16, activation='relu'))                 # 隐藏层，16个激活单元，激活函数为relu
model.add(Dense(units=10, activation='softmax'))              # 输出层，10个输出单元，激活函数为softmax

model.compile(loss='categorical_crossentropy', optimizer='sgd',metrics=['accuracy']) 

model.fit(X_train,y_train,batch_size=20,epochs= 50)

score_train = model.evaluate(X_train,y_train)
score_test = model.evaluate(X_test,y_test)

print('训练集loss值为：',score_train[0])
print('训练集准确率为：',score_train[1])

print('测试集loss值为：',score_test[0])
print('测试集准确率为：',score_test[1])