15、手写数字识别-小数据集

1.手写数字数据集

• from sklearn.datasets import load_digits
• digits = load_digits()

2.图片数据预处理

• x：归一化MinMaxScaler()
• y：独热编码OneHotEncoder()或to_categorical
• 训练集测试集划分
• 张量结构

3.设计卷积神经网络结构

• 绘制模型结构图，并说明设计依据。

4.模型训练

• model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
• train_history = model.fit(x=X_train,y=y_train,validation_split=0.2, batch_size=300,epochs=10,verbose=2)

5.模型评价

• model.evaluate()
• 交叉表与交叉矩阵
• pandas.crosstab
• seaborn.heatmap

from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import MinMaxScaler, OneHotEncoder
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns


def create_dataset():
    digits = load_digits()
    X_data = digits.data.astype(np.float32)
    Y_data = digits.target.astype(np.float32).reshape(-1, 1)  # 将Y_data变为一列

    return X_data, Y_data


def process_data(X_data, Y_data):
    # 将属性缩放到一个指定的最大和最小值（通常是1-0之间）
    scaler = MinMaxScaler()
    X_data = scaler.fit_transform(X_data)
    print("MinMaxScaler_trans_X_data:")
    print(X_data)
    Y = OneHotEncoder().fit_transform(Y_data).todense()  # 进行oe-hot编码
    print("one-hot_Y:")
    print(Y)
    return X_data, Y


def split_dataset(X_data, Y):
    # 转换为图片的格式（batch, height, width, channels）
    X = X_data.reshape(-1, 8, 8, 1)
    X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.2, random_state=0, stratify=Y)
    print(X_train.shape, X_test.shape, y_train.shape, y_test.shape)
    return X_train, X_test, y_train, y_test


def digits_model(X_train):
    """ 构建模型 """
    model = Sequential()

    ks = (3, 3)
    input_shape = X_train.shape[1:]

    # 一层卷积
    model.add(Conv2D(filters=64, kernel_size=ks, padding='same', input_shape=input_shape, activation='relu'))
    # 池化层1
    model.add(MaxPool2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))
    # 二层卷积
    model.add(Conv2D(filters=64, kernel_size=ks, padding='same', activation='relu'))
    # 池化层2
    model.add(MaxPool2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))

    model.add(Conv2D(filters=64, kernel_size=ks, padding='same', activation='relu'))
    model.add(Conv2D(filters=128, kernel_size=ks, padding='same', activation='relu'))
    model.add(MaxPool2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))

    model.add(Flatten())  # 平坦层
    model.add(Dense(128, activation='relu'))  # 全连接层
    model.add(Dropout(0.2))
    model.add(Dense(10, activation='softmax'))  # 激活函数

    print(model.summary())

    return model


def train_model(model, X_train, y_train):
    """ 训练模型 """
    model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
    train_history = model.fit(x=X_train, y=y_train, validation_split=0.2, batch_size=256, epochs=10, verbose=2)
    return train_history


def score_model(model, X_test, y_test):
    """ 评估模型 """
    return print(model.evaluate(X_test, y_test)[1])


def test_model(model, X_test):
    """ 测试模型 """
    y_pre = model.predict_classes(X_test)
    return y_pre


def crossrtab_matrix(y_test, y_pre):
    """
    交叉表、交叉矩阵
    查看预测数据与原数据对比
    """
    y_test = np.argmax(y_test, axis=1).reshape(-1)
    y_test = np.array(y_test)[0]
    # print(y_test)
    # print(type(y_test))
    # print('================')
    # print(type(y_pre))
    crosstab = pd.crosstab(y_test, y_pre, rownames=['labels'], colnames=['predict'])
    matrix = pd.DataFrame(crosstab)
    sns.heatmap(matrix, annot=True, cmap="RdPu", linewidths=0.2, linecolor='pink')
    plt.show()


def show_train_history(train_history, train, validation):
    plt.plot(train_history.history[train])
    plt.plot(train_history.history[validation])
    plt.title("Train History")
    plt.ylabel("train")
    plt.xlabel("epoch")
    plt.legend(['train', 'validation'], loc='upper left')
    plt.show()


if __name__ == '__main__':
    X_data, Y_data = create_dataset()
    X_data, Y = process_data(X_data, Y_data)
    X_train, X_test, y_train, y_test = split_dataset(X_data, Y)
    model = digits_model(X_train)
    train_history = train_model(model, X_train, y_train)
    score_model(model, X_test, y_test)
    y_pre = test_model(model, X_test)
    crossrtab_matrix(y_test, y_pre)
    show_train_history(train_history, 'accuracy', 'val_accuracy')  # 准确率
    show_train_history(train_history, 'loss', 'val_loss')  # 损失率

运行结果如下：

 

可视化展示数据训练参数最佳结果：

 

 　　由上图可知，当轮数epochs为10时，损失率以及准确率达到最佳。

交叉矩阵结果: