Summary of Non-myopic Multi-objective Bayesian Optimization, by Syrine Belakaria et al.
Non-Myopic Multi-Objective Bayesian Optimization
by Syrine Belakaria, Alaleh Ahmadianshalchi, Barbara Engelhardt, Stefano Ermon, Janardhan Rao Doppa
First submitted to arxiv on: 11 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary We investigate the problem of finite-horizon sequential experimental design to solve multi-objective optimization (MOO) of expensive black-box objective functions, a challenge that arises in applications like materials design. We propose non-myopic methods for MOO problems using Bayesian optimization (BO), addressing limitations in prior work on single-objective BO by introducing hypervolume improvement (HVI) as a scalarization approach. This allows us to develop three non-myopic acquisition functions (AFs) for MOO: Nested AF, Joint AF, and BINOM AF. Our experiments demonstrate that these AFs significantly improve performance over myopic AFs in multiple real-world MOO problems. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine trying to find the perfect material by testing many different combinations, but you only have a limited budget. This is like solving a multi-objective problem, where you want to find the best solution that satisfies several criteria. In this paper, scientists developed new methods to solve these kinds of problems using a technique called Bayesian optimization. They used an approach called hypervolume improvement to make sure their method was effective. The team tested their methods on many different real-world problems and found that they worked much better than other methods. |
Keywords
» Artificial intelligence » Optimization
