AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

IMMA: Immunizing text-to-image Models against Malicious Adaptation

Amber Yijia Zheng, Raymond A. Yeh

ECCV 2024 · Image generation › Text-to-image diffusion · approach: Text generation and structured prediction · 6 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
immunize the ESD-erased model with IMMA before LoRA re-learningESD erased SD V1-4 · Object re-learning (10 ImageNet classes)accuracy of target class (%): 45.5 -> 2.2Improved
immunize the model with IMMA (effect on other, non-target concepts)ESD erased SD V1-4 · Object re-learning (10 ImageNet classes)accuracy of other objects (%): 70.6 -> 49.3Hurt
apply IMMA against re-learning of artist stylesESD erased SD V1-4 · Artistic style re-learning (8 artists)SGR (LPIPS, CLIP, DINO): 21.84%, 5.5%, 23.35%Improved
apply IMMA against Textual Inversion personalizationSD V1-4 · CustomConcept101 (10 concepts)SGR(D) (%): 36.55 averageImproved
apply IMMA against DreamBooth personalizationSD V1-4 · CustomConcept101 (chair, lighthouse)SGR(C): -8.60 (chair), -1.19 (lighthouse); average 2.27Mixed
apply IMMA against NSFW nudity re-learningnudity-erased SD V1-4 · I2P unsafe promptschange in detected nudity (%): -60% -> -80%Improved
include only upper-level maximization (direct maximization) without lower-level adaptation (ablation)IMMA · DreamBooth personalization (woodenpot -> purse)image quality on other conceptHurt
add lower-level re-initialization from previous outer-loop phi (line 5) (ablation)IMMA · DreamBooth personalizationCLIP similarity to targetImproved
update overlapping parameters with adapted values (line 7) (ablation)IMMA · DreamBooth personalizationCLIP similarity to targetImproved
immunize against one adaptation method and adapt with another (crossed adaptation) (ablation)IMMA · DreamBooth and Textual Inversionqualitative target concept generationImproved
include JPEG-compressed images in IMMA training dataIMMA vs MIST · Textual Inversion with JPEG compressionqualitative protectionImproved
run IMMA sequentially on multiple target conceptsIMMA · castle, chair, guitar with Textual Inversionprotection and other-concept similarityMixed

Later papers that compared with this method

1 later accepted paper reports IMMA as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy