给定以下代码:
import matplotlib.pyplot as pltimport numpyfrom keras import callbacksfrom keras import optimizersfrom keras.layers import Dense, Dropoutfrom keras.models import Sequentialfrom keras.callbacks import ModelCheckpointfrom sklearn.preprocessing import StandardScalerfrom sklearn.ensemble import ExtraTreesClassifierfrom sklearn.utils import shuffle# Early stopping - 在过拟合前停止训练early_stop = callbacks.EarlyStopping(monitor='val_loss', min_delta=0, patience=3, verbose=1, mode='auto')# 为了可重复性固定随机种子seed = 42numpy.random.seed(seed)# 加载pima印第安人数据集dataset = numpy.loadtxt("./data/poc.csv",skiprows=1, delimiter=",")# 分割为输入(X)和输出(Y)变量X = dataset[:, 0:14]Y = dataset[:, 14:18]# # 通过移除均值和缩放到单位方差来标准化特征scaler = StandardScaler()X = scaler.fit_transform(X)# ADAM优化器带学习率衰减opt = optimizers.Adam(lr=0.0001, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0001)## 创建我们的模型model = Sequential()model.add(Dense(200, input_dim=14, kernel_initializer='uniform', activation='relu'))model.add(Dropout(0.2))model.add(Dense(100, activation='relu'))model.add(Dropout(0.2))model.add(Dense(60, activation='relu'))model.add(Dropout(0.2))model.add(Dense(30, activation='relu'))model.add(Dropout(0.2))model.add(Dense(5, activation='sigmoid'))model.summary()# 使用二元交叉熵编译模型，因为我们在预测0/1model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])# 检查点filepath="./checkpoints/weights.best.hdf5"checkpoint = ModelCheckpoint(filepath, monitor='val_acc', verbose=1, save_best_only=True, mode='max')# 拟合模型history = model.fit(X, Y, validation_split=0.33, epochs=10000, batch_size=10, verbose=0, callbacks=[early_stop,checkpoint])数据如下:
17.6,1,1,0,1,0,0,0,0,0,0,3.9,9.2,20.29,0,1,0,0,012.9,1,0,1,0,0,0,0,0,0,0,4.1,13.5,0.08,0,0,0,1,03.2,1,0,1,0,0,0,0,0,0,0,4.122031746,13.8,0.01,0,0,0,0,0...我得到了以下输出/错误:
_________________________________________________________________Layer (type) Output Shape Param # =================================================================dense_1 (Dense) (None, 200) 3000 _________________________________________________________________dropout_1 (Dropout) (None, 200) 0 _________________________________________________________________dense_2 (Dense) (None, 100) 20100 _________________________________________________________________dropout_2 (Dropout) (None, 100) 0 _________________________________________________________________dense_3 (Dense) (None, 60) 6060 _________________________________________________________________dropout_3 (Dropout) (None, 60) 0 _________________________________________________________________dense_4 (Dense) (None, 30) 1830 _________________________________________________________________dropout_4 (Dropout) (None, 30) 0 _________________________________________________________________dense_5 (Dense) (None, 5) 155 =================================================================Total params: 31,145Trainable params: 31,145Non-trainable params: 0_________________________________________________________________检查目标时出错: 期望dense_5的形状为(None, 1)，但得到的数组形状为(716, 4)
我遗漏了什么?
回答：
你的最后一层dense_5大小为5，而你的目标大小为4。
为了使其工作，每个目标的大小必须是你想要预测的类别数量。请记住，它们必须以独热编码(one hot encoding)的形式表示。你可以使用Keras的to_categorical函数来实现。
