Is there a way to combine different loss functions for a Neural Network?

Question

I am creating a NN that is outputting the percentage of each material (out of x number of materials) present in an object, so output is an (x, 1) array displaying the percentage