From 8f125a4922c947445ba3a987dd975417e638c902 Mon Sep 17 00:00:00 2001 From: minjong Date: Sun, 15 Sep 2024 05:02:06 +0900 Subject: [PATCH] feat: add research field of machine unlearning of t2i --- research_participate.html | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/research_participate.html b/research_participate.html index ecc18fc..896c0fd 100644 --- a/research_participate.html +++ b/research_participate.html @@ -92,6 +92,26 @@

# Machine Unlearning - Advisors: Jungseul Ok / Sangdon Park < +
  • +
    +
    + +
    +
    +

    # Machine Unlearning in Text-to-Image Generative Models - Advisor: Dongwoo Kim, Mentor: Saemi Moon, Minjong Lee

    +

    + Generative models have made significant advancements in producing high-quality images, but they also pose risks by potentially generating harmful, explicit, or privacy- and copyright-infringing content. To address these concerns, we aim to develop methods that allow models to “unlearn” problematic concepts, thereby enhancing the ethical and safe deployment of these models. +

    + References +
      +
    1. Moon, Saemi, Seunghyuk Cho, and Dongwoo Kim. "Feature Unlearning for Pre-trained GANs and VAEs." AAAI 2024
      2. +
    2. Kumari, Nupur, et al. "Ablating concepts in text-to-image diffusion models." NuerIPS 2023.
      3. +
    3. Gandikota, Rohit, et al. "Erasing Concepts from Diffusion Models." ICCV 2023.
      4. +
    +
    +
    +
    • +