I don't know about this specific case, but often the exact models are so complex that you can't optimize them analytically, so you resort to a global optimizati

I don't know about this specific case, but often the exact models are so complex that you can't optimize them analytically, so you resort to a global optimization best effort, where the results are not guaranteed to be any good - so you might as well try something DL-based too, to see how it would perform.