机器学习之手写数字识别-小数据集
Posted rakers1024
tags:
篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了机器学习之手写数字识别-小数据集相关的知识,希望对你有一定的参考价值。
1.手写数字数据集
# 导入手写数据集 from sklearn.datasets import load_digits data = load_digits() print(data)
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
""" @author Rakers """ import numpy as np # 导入手写数据集 from sklearn.datasets import load_digits # 图片数据预处理 --归一化 from sklearn.preprocessing import MinMaxScaler # OneHotEncoder独热编码 from sklearn.preprocessing import OneHotEncoder # 切分数据集 from sklearn.model_selection import train_test_split data = load_digits() # x:归一化MinMaxScaler() X_data = data[‘data‘].astype(np.float32) scaler = MinMaxScaler() X_data = scaler.fit_transform(X_data) print("归一化后数据: ",X_data) # 转化为图片的格式 X=X_data.reshape(-1, 8, 8, 1) print("转化为图片后数据:", X.shape) # y:独热编码OneHotEncoder() y = data[‘target‘].astype(np.float32).reshape(-1, 1) # 将Y_data变为一列 Y = OneHotEncoder().fit_transform(y).todense() # 张量结构todense print("Y独热编码: ", Y) 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,X_test,y_train,y_test) print("X_data.shape:",X_data.shape) print("X.shape:",X.shape)
3.设计卷积神经网络结构
绘制模型结构图,设计依据。
""" @author Rakers """ import numpy as np # 导入手写数据集 from sklearn.datasets import load_digits # 图片数据预处理 --归一化 from sklearn.preprocessing import MinMaxScaler # OneHotEncoder独热编码 from sklearn.preprocessing import OneHotEncoder # 切分数据集 from sklearn.model_selection import train_test_split from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense,Dropout,Conv2D,MaxPool2D,Flatten def buildModel(isPrintSummary=True, X_train=None): """ # 建立模型 :param isPrintSummary: 是否打印Summary信息 :return: 返回构建的模型 """ model = Sequential() ks = (3, 3) # 卷积核的大小 input_shape = X_train.shape[1:] # 一层卷积,padding=‘same‘,tensorflow会对输入自动补0 model.add(Conv2D(filters=16, kernel_size=ks, padding=‘same‘, input_shape=input_shape, activation=‘relu‘)) # 池化层1 model.add(MaxPool2D(pool_size=(2, 2))) # 防止过拟合,随机丢掉连接 model.add(Dropout(0.25)) # 二层卷积 model.add(Conv2D(filters=32, kernel_size=ks, padding=‘same‘, activation=‘relu‘)) # 池化层2 model.add(MaxPool2D(pool_size=(2, 2))) model.add(Dropout(0.25)) # 三层卷积 model.add(Conv2D(filters=64, kernel_size=ks, padding=‘same‘, activation=‘relu‘)) # 四层卷积 model.add(Conv2D(filters=128, kernel_size=ks, padding=‘same‘, activation=‘relu‘)) # 池化层3 model.add(MaxPool2D(pool_size=(2, 2))) model.add(Dropout(0.25)) # 平坦层 model.add(Flatten()) # 全连接层 model.add(Dense(128, activation=‘relu‘)) model.add(Dropout(0.25)) # 激活函数softmax model.add(Dense(10, activation=‘softmax‘)) if isPrintSummary: print(model.summary()) return model if __name__ == "__main__": data = load_digits() # x:归一化MinMaxScaler() X_data = data[‘data‘].astype(np.float32) scaler = MinMaxScaler() X_data = scaler.fit_transform(X_data) # print("归一化后数据: ", X_data) # 转化为图片的格式 X = X_data.reshape(-1, 8, 8, 1) # print("转化为图片后数据:", X.shape) # y:独热编码OneHotEncoder() y = data[‘target‘].astype(np.float32).reshape(-1, 1) # 将Y_data变为一列 Y = OneHotEncoder().fit_transform(y).todense() # 张量结构todense # print("Y独热编码: ", Y) 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, X_test, y_train, y_test) # print("X_data.shape:", X_data.shape) # print("X.shape:", X.shape) model = buildModel(X_train=X_train)
4.模型训练
""" @author Rakers """ import numpy as np import matplotlib.pyplot as plt # 导入手写数据集 from sklearn.datasets import load_digits # 图片数据预处理 --归一化 from sklearn.preprocessing import MinMaxScaler # OneHotEncoder独热编码 from sklearn.preprocessing import OneHotEncoder # 切分数据集 from sklearn.model_selection import train_test_split from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense,Dropout,Conv2D,MaxPool2D,Flatten def buildModel(isPrintSummary=True, X_train=None): """ # 建立模型 :param isPrintSummary: 是否打印Summary信息 :return: 返回构建的模型 """ model = Sequential() ks = (3, 3) # 卷积核的大小 input_shape = X_train.shape[1:] # 一层卷积,padding=‘same‘,tensorflow会对输入自动补0 model.add(Conv2D(filters=16, kernel_size=ks, padding=‘same‘, input_shape=input_shape, activation=‘relu‘)) # 池化层1 model.add(MaxPool2D(pool_size=(2, 2))) # 防止过拟合,随机丢掉连接 model.add(Dropout(0.25)) # 二层卷积 model.add(Conv2D(filters=32, kernel_size=ks, padding=‘same‘, activation=‘relu‘)) # 池化层2 model.add(MaxPool2D(pool_size=(2, 2))) model.add(Dropout(0.25)) # 三层卷积 model.add(Conv2D(filters=64, kernel_size=ks, padding=‘same‘, activation=‘relu‘)) # 四层卷积 model.add(Conv2D(filters=128, kernel_size=ks, padding=‘same‘, activation=‘relu‘)) # 池化层3 model.add(MaxPool2D(pool_size=(2, 2))) model.add(Dropout(0.25)) # 平坦层 model.add(Flatten()) # 全连接层 model.add(Dense(128, activation=‘relu‘)) model.add(Dropout(0.25)) # 激活函数softmax model.add(Dense(10, activation=‘softmax‘)) if isPrintSummary: print(model.summary()) return model # 画Train History图 def show_train_history(train_history, train, validation): """ @author Rakers :param train_history: :param train: :param validation: :return: """ if train in train_history.history: plt.plot(train_history.history[train]) if validation in train_history.history: 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__": data = load_digits() # x:归一化MinMaxScaler() X_data = data[‘data‘].astype(np.float32) scaler = MinMaxScaler() X_data = scaler.fit_transform(X_data) # print("归一化后数据: ", X_data) # 转化为图片的格式 X = X_data.reshape(-1, 8, 8, 1) # print("转化为图片后数据:", X.shape) # y:独热编码OneHotEncoder() y = data[‘target‘].astype(np.float32).reshape(-1, 1) # 将Y_data变为一列 Y = OneHotEncoder().fit_transform(y).todense() # 张量结构todense # print("Y独热编码: ", Y) 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, X_test, y_train, y_test) # print("X_data.shape:", X_data.shape) # print("X.shape:", X.shape) model = buildModel(X_train=X_train) # 模型训练 model.compile(loss=‘categorical_crossentropy‘, optimizer=‘adam‘, metrics=[‘acc‘]) train_history = model.fit(x=X_train, y=y_train, validation_split=0.2, batch_size=300, epochs=10, verbose=2) # 准确率 show_train_history(train_history, ‘acc‘, ‘val_acc‘) # 损失率 show_train_history(train_history, ‘loss‘, ‘val_loss‘)
5.模型评价
- model.evaluate()
- 交叉表与交叉矩阵
- pandas.crosstab
- seaborn.heatmap
""" @author Rakers """ import numpy as np import pandas as pd import seaborn as sns import matplotlib.pyplot as plt # 导入手写数据集 from sklearn.datasets import load_digits # 图片数据预处理 --归一化 from sklearn.preprocessing import MinMaxScaler # OneHotEncoder独热编码 from sklearn.preprocessing import OneHotEncoder # 切分数据集 from sklearn.model_selection import train_test_split from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense,Dropout,Conv2D,MaxPool2D,Flatten def buildModel(isPrintSummary=True, X_train=None): """ # 建立模型 :param isPrintSummary: 是否打印Summary信息 :return: 返回构建的模型 """ model = Sequential() ks = (3, 3) # 卷积核的大小 input_shape = X_train.shape[1:] # 一层卷积,padding=‘same‘,tensorflow会对输入自动补0 model.add(Conv2D(filters=16, kernel_size=ks, padding=‘same‘, input_shape=input_shape, activation=‘relu‘)) # 池化层1 model.add(MaxPool2D(pool_size=(2, 2))) # 防止过拟合,随机丢掉连接 model.add(Dropout(0.25)) # 二层卷积 model.add(Conv2D(filters=32, kernel_size=ks, padding=‘same‘, activation=‘relu‘)) # 池化层2 model.add(MaxPool2D(pool_size=(2, 2))) model.add(Dropout(0.25)) # 三层卷积 model.add(Conv2D(filters=64, kernel_size=ks, padding=‘same‘, activation=‘relu‘)) # 四层卷积 model.add(Conv2D(filters=128, kernel_size=ks, padding=‘same‘, activation=‘relu‘)) # 池化层3 model.add(MaxPool2D(pool_size=(2, 2))) model.add(Dropout(0.25)) # 平坦层 model.add(Flatten()) # 全连接层 model.add(Dense(128, activation=‘relu‘)) model.add(Dropout(0.25)) # 激活函数softmax model.add(Dense(10, activation=‘softmax‘)) if isPrintSummary: print(model.summary()) return model # 画Train History图 def show_train_history(train_history, train, validation): """ @author Rakers :param train_history: :param train: :param validation: :return: """ if train in train_history.history: plt.plot(train_history.history[train]) if validation in train_history.history: 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__": data = load_digits() # x:归一化MinMaxScaler() X_data = data[‘data‘].astype(np.float32) scaler = MinMaxScaler() X_data = scaler.fit_transform(X_data) # print("归一化后数据: ", X_data) # 转化为图片的格式 X = X_data.reshape(-1, 8, 8, 1) # print("转化为图片后数据:", X.shape) # y:独热编码OneHotEncoder() y = data[‘target‘].astype(np.float32).reshape(-1, 1) # 将Y_data变为一列 Y = OneHotEncoder().fit_transform(y).todense() # 张量结构todense # print("Y独热编码: ", Y) 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, X_test, y_train, y_test) # print("X_data.shape:", X_data.shape) # print("X.shape:", X.shape) model = buildModel(X_train=X_train) # 模型训练 model.compile(loss=‘categorical_crossentropy‘, optimizer=‘adam‘, metrics=[‘acc‘]) train_history = model.fit(x=X_train, y=y_train, validation_split=0.2, batch_size=300, epochs=10, verbose=2) # 准确率 show_train_history(train_history, ‘acc‘, ‘val_acc‘) # 损失率 show_train_history(train_history, ‘loss‘, ‘val_loss‘) # 模型评价 score = model.evaluate(X_test, y_test) print(‘score:‘, score) # 预测值 y_pred = model.predict_classes(X_test) print(‘y_pred:‘, y_pred[:10]) # 交叉表与交叉矩阵 y_test1 = np.argmax(y_test, axis=1).reshape(-1) y_true = np.array(y_test1)[0] # 交叉表查看预测数据与原数据对比 # pandas.crosstab pd.crosstab(y_true, y_pred, rownames=[‘true‘], colnames=[‘predict‘]) # 交叉矩阵 # seaborn.heatmap y_test1 = y_test1.tolist()[0] a = pd.crosstab(np.array(y_test1), y_pred, rownames=[‘Lables‘], colnames=[‘Predict‘]) # 转换成属dataframe df = pd.DataFrame(a) sns.heatmap(df, annot=True, cmap="Reds", linewidths=0.2, linecolor=‘G‘) plt.show()
以上是关于机器学习之手写数字识别-小数据集的主要内容,如果未能解决你的问题,请参考以下文章