compare NN and SVM approaches from the application point of view, their similarities and differences, how they each handle multicollinearity and unbalancing in the data, and what kind of measure of performance we should consider when we have an unbalanced data using any of these methods?

You should use at least 3-5 references.