Summary of Training Greedy Policy For Proposal Batch Selection in Expensive Multi-objective Combinatorial Optimization, by Deokjae Lee et al.
Training Greedy Policy for Proposal Batch Selection in Expensive Multi-Objective Combinatorial Optimization
by Deokjae Lee, Hyun Oh Song, Kyunghyun Cho
First submitted to arxiv on: 21 Jun 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 This paper presents a novel approach to subset selection for expensive multi-objective combinatorial optimization problems, specifically focusing on optimizing the batch acquisition score. The existing methods optimize individual acquisition scores or latent space-based approaches, but these have limitations in handling dependencies among candidates and large search spaces. To address this challenge, the authors introduce a greedy-style algorithm that optimizes batch acquisition directly on the combinatorial space through sequential sampling from a trained policy. This method is demonstrated to be efficient, achieving baseline performance in 1.69x fewer queries on the red fluorescent proteins design task. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper solves a tricky problem in machine learning called subset selection for big optimization tasks. It’s like finding the best combination of things to try next when you have many options and want to get the right answer quickly. Right now, people are using methods that work okay but aren’t perfect because they don’t take into account how different options might be related. The authors came up with a new way to do this by breaking down the big problem into smaller ones and solving them one at a time. This makes it much faster and more efficient! They tested it on designing proteins, which is a complex task that requires trying many combinations of things before you get the right answer. |
Keywords
» Artificial intelligence » Latent space » Machine learning » Optimization
