In Bayesian optimization, when I use the Multi-objective Expected Improvement (MEI) function, it tends to cluster samples in a specific zone of the objective space leaving other zones unexplored. I would like to have a more uniform expansion of the Pareto front. I have faced this issue so many times, in different optimization problems. I am not sure if that is a common behavior of the MEI or a problem in my code (bug). Have you faced this "clustering issue"?  What have you done to overcome it?