Summary of Data Attribution For Text-to-image Models by Unlearning Synthesized Images, By Sheng-yu Wang et al.
Data Attribution for Text-to-Image Models by Unlearning Synthesized Images
by Sheng-Yu Wang, Aaron Hertzmann, Alexei A. Efros, Jun-Yan Zhu, Richard Zhang
First submitted to arxiv on: 13 Jun 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: 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 proposed data attribution method for text-to-image models identifies the training images most influencing the generation of a new image by simulating unlearning the synthesized image. The approach increases the training loss on the output image without forgetting unrelated concepts, allowing for efficient identification of influential images. This method outperforms previous approaches in computationally intensive evaluations. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper proposes a new way to find which pictures a text-to-image model learned from most when generating a new picture. They do this by making the model “unlearn” what it knows about the new picture, without forgetting what it already knows. This helps them figure out which training images are most important for the generated image. The method is tested and shown to be better than existing methods. |
