Machine Learning (CNN): 识别手写数字

机器学习入门 Machine Learning Tutorial I: 图像 Images

1. 手写数字识别 Handwritten Digits Recognition

附上对应英文，有助于日后更深层次的学习；若对内容有任何疑问，欢迎留言讨论 : )

内容简介

通过训练简单的CNN模型 （Convolutional Neural Networks 卷积神经网络）来识别MNIST数据集中的手写数字，精确度高于99%。

环境准备

Python 3.6
pandas 0.24.2
tensorflow 1.13.1 / keras 2.2.4

1. 导入模块 (Import module)

import pandas as pd
from keras import datasets, layers, models # 或 from tensorflow.keras import datasets, layers, models

2. 下载及处理MNIST数据集 (Download and prepare the MNIST dataset)

数据来源：keras.datasets.minist
training data: 60000 images, 28x28 pixels for each image
testing data: 10000 images, 28x28 pixels for each image
图像示例：
后附数据视觉化(Data visualization)代码

image.png

下载数据集：

(train_images, train_labels), (test_images, test_labels) = datasets.mnist.load_data()

处理数据集：

train_images = train_images.reshape((60000, 28, 28, 1))
test_images = test_images.reshape((10000, 28, 28, 1))
train_images, test_images = train_images / 255.0, test_images / 255.0 # 将像素值标准化为至0-1区间 (原为0-255)

数据视觉化（如果你想看看图片）：

def visualize(X, label):
    # Convert train datset to (num_images, img_rows, img_cols) format
    X = X.reshape(X.shape[0], 28, 28)
    for i in range(0, 9):
        plt.subplot(330+(i+1))
        plt.imshow(X[i], cmap=plt.get_cmap('gray'))
        plt.title(label[i])
    plt.show()

visualize(train_images, train_labels)

3. 建立模型 (Build the model)

model = models.Sequential()
model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(28,28,1)))
model.add(layers.MaxPool2D(2, 2))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.MaxPool2D(2, 2))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.Flatten())
model.add(layers.Dense(64, activation='relu'))
model.add(layers.Dense(10, activation='softmax'))

4. 编译及训练模型 (Compile and train the model)

model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])
model.fit(train_images, train_labels, epochs=5)

5. 评估模型 (Evaluate the model)

test_loss, test_acc = model.evaluate(test_images, test_labels)
print("Test Accuracy: ", test_acc)

准确率 Accuracy: 99.16%

image.png

6. 输出预测结果 (Output the predictions)

predictions = model.predict_classes(test_images, verbose=0)
submissions=pd.DataFrame({"ImageId": list(range(1,len(predictions)+1)),
                         "Label": predictions})
submissions.to_csv("predictions_cnn.csv", index=False, header=True)

即可在当前文件夹中看到predictions_cnn.csv文件，内含预测结果。

学习自：https://www.tensorflow.org/beta/tutorials/images/intro_to_cnns