[P] Is there an equivalent of Bayesian optimization that works only with comparative results?

Hello everyone, I'm working on a problem where I need to find the best set of parameters (10 of them) that optimises a very costly objective function. Normally, I would use a Bayesian optimisation, but in this specific case, I don't have access to the actual objective function, the only thing that I can calculate is weather the function is higher with a certain set of parameters A or B. I don't know how the actual values of the function, …

bayesian case function hello machinelearning normally optimisation optimization parameters set them