Summary of Derivative-free Guidance in Continuous and Discrete Diffusion Models with Soft Value-based Decoding, by Xiner Li et al.
Derivative-Free Guidance in Continuous and Discrete Diffusion Models with Soft Value-Based Decoding
by Xiner Li, Yulai Zhao, Chenyu Wang, Gabriele Scalia, Gokcen Eraslan, Surag Nair, Tommaso Biancalani, Shuiwang Ji, Aviv Regev, Sergey Levine, Masatoshi Uehara
First submitted to arxiv on: 15 Aug 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Genomics (q-bio.GN); Machine Learning (stat.ML)
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 proposes a new method for optimizing downstream reward functions while preserving the naturalness of design spaces in various domains, including images, molecules, DNA/RNA sequences, and protein sequences. The proposed algorithm integrates soft value functions into the standard inference procedure of pre-trained diffusion models, allowing for iterative sampling that looks ahead to future rewards. This approach avoids fine-tuning generative models and eliminates the need for differentiable proxy models. The method is demonstrated across several domains, including image generation, molecule generation, and DNA/RNA sequence generation. The paper’s contributions include direct utilization of non-differentiable features/reward feedback and principled application to recent discrete diffusion models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research aims to make it easier to generate natural-looking images, molecules, and DNA sequences that also meet specific goals or criteria. Right now, generating these things requires using complex models that can be difficult to work with. The authors of this paper developed a new way to do this that’s more straightforward and doesn’t require as much setup. They tested their method on various tasks, including creating realistic images and designing molecules, and showed it works well. |
Keywords
» Artificial intelligence » Fine tuning » Image generation » Inference
