Summary of Conceptprune: Concept Editing in Diffusion Models Via Skilled Neuron Pruning, by Ruchika Chavhan and Da Li and Timothy Hospedales
ConceptPrune: Concept Editing in Diffusion Models via Skilled Neuron Pruning
by Ruchika Chavhan, Da Li, Timothy Hospedales
First submitted to arxiv on: 29 May 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 presents a novel approach to unlearning undesirable concepts from pre-trained text-to-image generation models without requiring additional training or data. The proposed method, ConceptPrune, identifies critical regions within the model responsible for generating specific concepts and enables efficient erasure of these concepts via weight pruning. The approach is shown to be effective in removing artistic styles, nudity, objects, and gender biases from generated images while also being robust against various adversarial attacks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper introduces a way to remove unwanted ideas from big image-generating models without needing more training or data. This method, called ConceptPrune, finds the parts of the model that create specific concepts and lets you erase them easily by pruning certain weights. The approach works well in removing artistic styles, nudity, objects, and gender biases from generated images. |
Keywords
» Artificial intelligence » Image generation » Pruning
