如何在python（keras）中进行反向预测？

我有一个分类神经网络，我在一个非常大的数据集中训练，具有相当的准确性和预测性。现在我想反向预测（正常预测：input->output反向预测input

from keras.models import Sequential
from keras.layers import Dense
import numpy as np

#model
model = Sequential()
model.add(Dense(1,input_dim=1))
model.add(Dense(4))
model.add(Dense(4))
model.add(Dense(2, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy'])
model.summary

#creating natural number features and binary odd and even targets
n=500
#natural numbers
x=np.arange(1,n)
x=x.reshape(n-1,1)
#even
y1=np.where(x%2==0,1,0)
#odd
y2=np.where(x%2!=0,1,0)

#target
Y=np.column_stack([y1,y2])

#model training
model.fit(x,Y,validation_split=0.3)

y_dash=np.arange(n,n+10)

#model prediction
model.predict(y_dash)

349/349 [==============================] - 0s - loss: 0.7077 - acc: 0.4871 - val_loss: 0.6991 - val_acc: 0.5000