Many real world applications can be framed as multi-objective optimization, where we wish to simultaneously optimize for multiple criteria. optimization techniques for the multi-objective setting are pertinent the evaluation of the functions in question are expensive. Traditional for multi-objective optimization, both Bayesian and otherwise, are at recovering the Pareto front of these objectives. However, in certain a practitioner might desire to identify Pareto optimal points only in a of the Pareto front due to external considerations. In this work, we a strategy based on random scalarizations of the objectives that this problem. Our approach is able to flexibly sample from desired of the Pareto front and, computationally, is considerably cheaper than approaches for MOO. We also study a notion of regret in the-objective setting and show that our strategy achieves sublinear regret. experiment with both synthetic and real-life problems, and demonstrate performance of our proposed algorithm in terms of the flexibility and.
No takes yet. Share an insight, caveat, or question.
Paria et al. (2018) studied this question.