ValueError: Shapes (None, 1) and (None, 90) are incompatible

Question

我想在我的 x_train 和我的 y_train 中构建一个 deep RNN。当我执行下面的代码时：

print(X_train_fea.shape, y_train_fea.shape)
X_train_res = np.reshape(X_train_fea,(10510,10,1))
y_train_res = np.reshape(y_train_fea.to_numpy(),(-1,1))
print(X_train_res.shape, y_train_res.shape)

结果：

(10510, 10) (10510,)
(10510, 10, 1) (10510, 1)

和

model = Sequential([
    LSTM(90, input_shape=(10,1)), 
])
model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])
model.summary()

当我拟合模型时

history = model.fit(X_train_res, y_train_res,epochs=5)

我得到了

ValueError: Shapes (None, 1) and (None, 90) are incompatible

Answer 1

看起来 y_train_res 包含整数索引而不是单热向量。如果是这样，您必须使用 sparse_categorical_crossentropy:

model.compile(optimizer='adam', loss='sparse_categorical_crossentropy')

并将其形状更改为一维：

y_train_res = np.reshape(y_train_fea.to_numpy(),(-1,))

ValueError: Shapes (None, 1) and (None, 90) are incompatible

ValueError: Shapes (None, 1) and (None, 90) are incompatible

python-3.x

lstm

tensorflow