使用keras在深度学习模型中训练数据形状错误

Question

我正在为多类分类问题创建深度学习模型，我的模型包含 46 个唯一类。我的 X_train 形状是 (14382, 183), & y_train是 14382

代码-

#Creating Dummy Variables 
X=pd.get_dummies(X, prefix=list((X.select_dtypes(include=[object])).columns))

#Splitting the dataset
from sklearn.model_selection import train_test_split
X_train, X_valid, y_train, y_valid= train_test_split(X, y, test_size=0.3, random_state=10)

model = Sequential()

#adding layers to the model
model.add(Dense(units =367, activation ='relu', input_dim =183))
model.add(Dense(units =182, activation ='relu'))
model.add(Dense(units =182, activation='relu'))


 #output layer
model.add(Dense(46, activation='softmax'))
model.compile(loss = 'categorical_crossentropy' , optimizer = keras.optimizers.Adam(learning_rate=0.0001) , metrics = ['accuracy'] )
model.fit(X_train, y_train, epochs=20, batch_size = 50, validation_data=(X_valid, y_valid))

我在运行模型之后遇到错误-

ValueError: Error when checking target: expected dense_110 to have shape (46,) but got array with shape (1,)

如何解决这个错误？

Answer 1

问题出在你的目标形状上

实际上它是一维的，所以你可以保持原样并应用 sparse_categorical_crossentropy 作为损失函数

X = np.random.randint(0,10, (1000,100))
y = np.random.randint(0,3, 1000)

model = Sequential([
    Dense(128, input_dim = 100),
    Dense(3, activation='softmax'),
])
model.summary()
model.compile(loss='sparse_categorical_crossentropy',optimizer='adam',metrics=['accuracy'])
history = model.fit(X, y, epochs=3)

否则，您可以对其进行一次性编码 pd.get_dummies(y).values（在训练测试拆分之前）并获得形状为 (n_sample、n_class) 的 y。在这种情况下，您可以使用 categorical_crossentropy

X = np.random.randint(0,10, (1000,100))
y = pd.get_dummies(np.random.randint(0,3, 1000)).values

model = Sequential([
    Dense(128, input_dim = 100),
    Dense(3, activation='softmax'),
])
model.summary()
model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy'])
history = model.fit(X, y, epochs=3)

使用keras在深度学习模型中训练数据形状错误

train data shape error in the deep learning model using keras

machine-learning

neural-network

deep-learning

keras

tf.keras