Summary of Unstable Unlearning: the Hidden Risk Of Concept Resurgence in Diffusion Models, by Vinith M. Suriyakumar et al.
Unstable Unlearning: The Hidden Risk of Concept Resurgence in Diffusion Models
by Vinith M. Suriyakumar, Rohan Alur, Ayush Sekhari, Manish Raghavan, Ashia C. Wilson
First submitted to arxiv on: 10 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV)
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 explores a vulnerability in text-to-image diffusion models when fine-tuning them on seemingly unrelated images, which can cause previously “unlearned” concepts to be relearned. The study uses Stable Diffusion v1.4 and v2.1 models to investigate this phenomenon, termed concept resurgence. The findings highlight the fragility of incremental model updates and raise concerns about ensuring the safety and alignment of these models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper shows that text-to-image diffusion models can learn old concepts again even when trying to forget them. When updating an existing model with new images, it’s possible for the model to remember things it was previously told to forget. This is a problem because it means that just making small changes to a model can cause it to recall unwanted ideas. The researchers tested this by fine-tuning two models and found that they were able to relearn old concepts. |
Keywords
» Artificial intelligence » Alignment » Diffusion » Fine tuning » Recall
