Summary of Moesd: Mixture Of Experts Stable Diffusion to Mitigate Gender Bias, by Guorun Wang et al.
MoESD: Mixture of Experts Stable Diffusion to Mitigate Gender Bias
by Guorun Wang, Lucia Specia
First submitted to arxiv on: 25 Jun 2024
Categories
- Main: Computation and Language (cs.CL)
- 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 proposed Mixture-of-Experts (MoESD) with BiAs (Bias Adapters) model aims to mitigate gender bias in text-to-image models by identifying and addressing biases in the latent space. The method, which builds upon a stable diffusion framework, introduces a Bias-Identification Gate mechanism to detect and correct social biases present in the text encoder of the model. Experimental results demonstrate the effectiveness of MoESD with BiAs in reducing gender bias while preserving image quality. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new approach is being developed to make AI models less biased against certain genders or ethnicities. The current AI models can create images that are unfair and stereotypical, such as showing mostly men in leadership positions. To fix this issue, the researchers propose a new method called MoESD with BiAs. This method helps identify biases in AI models and correct them to make sure they’re fairer and more inclusive. By adding a special token to the prompt, the model can learn to reduce gender bias without sacrificing image quality. |
Keywords
» Artificial intelligence » Diffusion » Encoder » Latent space » Mixture of experts » Prompt » Token
