So the question basically is; does it matter what the inputs are of a MLP NN in the sense of input dependence. For example:
Input 1: A
Input 2: B
Input 3: A+B or A*B or any other function using A and B as variables.
For more basic regression models this sometimes makes a huge impact in how we interpret results or the functionality of the model in general. This lead me to question what happens in a general MLP NN.