Summary of Dmin: Scalable Training Data Influence Estimation For Diffusion Models, by Huawei Lin et al.
DMin: Scalable Training Data Influence Estimation for Diffusion Models
by Huawei Lin, Yingjie Lao, Weijie Zhao
First submitted to arxiv on: 11 Dec 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
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 paper proposes a scalable framework called DMin for estimating the influence of each training data sample on generated images from diffusion models (DMs). Existing methods are limited to small-scale or LoRA-tuned models due to computational constraints. DMin leverages efficient gradient compression, reducing storage requirements and allowing for fast retrieval of influential samples. The method can handle DMs with billions of parameters, making it the first to achieve influence estimation at this scale. The authors demonstrate the effectiveness and efficiency of DMin through empirical results. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine you’re trying to make a beautiful painting using AI algorithms. You want to know which parts of the original pictures helped create the final artwork. This paper introduces a new way to figure out which training images are most important for making these generated images. The method, called DMin, is special because it can handle massive amounts of data and still work fast. The researchers tested DMin and showed that it’s both good at finding influential images and quick to use. |
Keywords
» Artificial intelligence » Diffusion » Lora
