Multiple losses for imbalanced dataset with Keras

Question

My Model:

I built a siamese network that take two input and has three outputs. So My loss functions is :

total loss = alpha( loss1)