Keras custom predicted value by adding new weight parameter (custom loss function/ add new lambda layer)

Question

I am trying to construct my predicted value as:

y_hat = alpha * output_1 + (1-alpha) * output_2

And the "output_1" and "output