English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Advancements in text-to-image models and fine-tuning methods have led to theincreasing risk of malicious adaptation, i.e., fine-tuning to generate harmfulunauthorized content. Recent works, e.g., Glaze or MIST, have developeddata-poisoning techniques which protect the data against adaptation methods. Inthis work, we consider an alternative paradigm for protection. We propose to``immunize'' the model by learning model parameters that are difficult for theadaptation methods when fine-tuning malicious content; in short IMMA. Empiricalresults show IMMA's effectiveness against malicious adaptations, includingmimicking the artistic style and learning of inappropriate/unauthorizedcontent, over three adaptation methods: LoRA, Textual-Inversion, andDreamBooth.

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