Keras_CNN网络

# -*- coding: utf-8 -*- """ Created on Fri Aug 3 18:24:15 2018 @author: leodflag """ import keras from keras.datasets import mnist from keras.models import Sequential from keras.layers import Dense, Dropout, Flatten from keras.layers import Conv2D, MaxPooling2D # mnist 是灰度手寫數字圖片資料庫 # batch_size是整數，指定进行梯度下降时每个batch包含的样本数 batch_size = 128 # 10個類別標籤 0-9 num_classes = 10 epochs = 1 """ 將圖片預處理 """ # 圖片28*28的尺寸 img_rows, img_cols=28, 28 # 下載資料 # X shape (60,000 28x28), y shape (10,000, ) (x_train, y_train), (x_test, y_test)= mnist.load_data() # reshape：将tensor变换为参数shape的形式，shape是列表形式 x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1) x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1) # 28*28大小的黑白圖片 input_shape = (img_rows, img_cols, 1) # astype 轉換數據類型 x_train = x_train.astype('float32') x_test = x_test.astype('float32') # 標準化，從0-255變成0-1之間 x_train /= 255 x_test /= 255 print('x_train shape:', x_train.shape) print(x_train.shape[0], 'train samples') print(x_test.shape[0], 'test samples') # 將0-9之間的y改造成一個大小為10的向量，屬於哪個數字， # 就在哪個位置為1，其他為0 y_train = keras.utils.to_categorical(y_train, num_classes) y_test = keras.utils.to_categorical(y_test, num_classes) """ 建立神經網路 """ # 建立神經層 model = Sequential() # 32個3*3filters 、relu 激活函數、28*28*1的輸入，圖片28*28尺寸,1表黑白 # Conv2D，2D空間卷積層 model.add(Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=input_shape)) # 64個3*3filters、relu 激活函數、input_shape為Conv2D在第一層使用時才用 model.add(Conv2D(64, (3, 3), activation='relu')) # 用2*2的最大池化矩陣去壓縮資料 model.add(MaxPooling2D(pool_size=(2, 2))) # Dropout在训练中每次更新时， # 可隨機刪除部分神經元，防止過擬合 model.add(Dropout(0.25)) # Flatten，建立平坦層，將64個12*12的矩陣，轉換成一維向量9216個神經元 model.add(Flatten()) # Dense，神經元連接層，128個神經元，激活函數：relu model.add(Dense(128, activation='relu')) # 刪除部分神經元 model.add(Dropout(0.5)) # Dense，輸出層，10個神經元，輸出維度10的矩陣，激活函數：softmax model.add(Dense(num_classes, activation='softmax')) # compile編譯模型以提供訓練 # loss：損失函數 # optimizer：優化器 model.compile(loss=keras.losses.categorical_crossentropy, optimizer=keras.optimizers.Adadelta(), metrics=['accuracy']) # fit 訓練模型 model.fit(x_train, y_train, batch_size=batch_size, epochs=epochs, verbose=1, validation_data=(x_test, y_test)) # 返回一个测试误差的标量值，或一个标量的list score = model.evaluate(x_test, y_test, verbose=0) print('Test loss:', score[0]) print('Test accuracy:', score[1])