Summary of Robust Guided Diffusion For Offline Black-box Optimization, by Can Sam Chen et al.
Robust Guided Diffusion for Offline Black-Box Optimization
by Can Sam Chen, Christopher Beckham, Zixuan Liu, Xue Liu, Christopher Pal
First submitted to arxiv on: 1 Oct 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 The proposed approach, Robust Guided Diffusion (RGD), combines the advantages of proxy and proxy-free diffusion for effective conditional generation in offline black-box optimization. The forward approach learns a mapping from input to its value, while the inverse approach learns a mapping from value to input for conditional generation. However, proxy-free diffusion lacks explicit guidance from proxies, which is essential for generating high-performance samples beyond the training distribution. To address this, RGD utilizes proxy-enhanced sampling, which utilizes the explicit guidance from a trained proxy to bolster proxy-free diffusion with enhanced sampling control. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Offline black-box optimization aims to maximize a black-box function using an offline dataset of designs and their measured properties. The approach combines two main methods: forward and inverse approaches. While both have benefits, they also have limitations. For example, the forward approach lacks explicit guidance from proxies, which is essential for generating high-performance samples beyond the training distribution. To overcome this limitation, RGD uses proxy-enhanced sampling to boost proxy-free diffusion with enhanced sampling control. |
Keywords
» Artificial intelligence » Diffusion » Optimization
