EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

AI model disgorgement: Methods and choices

Alessandro Achille, Michael Kearns, Carson Klingenberg iD, Stefano Soatto

Proceedings of the National Academy of Sciences · 30.04.2024 · 121 (18)

DOI: 10.1073/pnas.2307304121 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the National Academy of Sciences, online erschienen am 19.04.2024. Verfasst von Alessandro Achille und 3 weiteren Autor:innen. Beteiligte Einrichtung: Amazon Web Services Artificial Intelligence (AWS AI), Pasadena, CA 91125. Hauptthemen laut MEDLINE: Language, Machine Learning. Die Arbeit wurde bisher 9-mal zitiert (Crossref) und nennt 48 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Over the past few years, machine learning models have significantly increased in size and complexity, especially in the area of generative AI such as large language models. These models require massive amounts of data and compute capacity to train, to the extent that concerns over the training data (such as protected or private content) cannot be practically addressed by retraining the model “from scratch” with the questionable data removed or altered. Furthermore, despite significant efforts and controls dedicated to ensuring that training corpora are properly curated and composed, the sheer volume required makes it infeasible to manually inspect each datum comprising a training corpus. One potential approach to training corpus data defects is model disgorgement, by which we broadly mean the elimination or reduction of not only any improperly used data, but also the effects of improperly used data on any component of an ML model. Model disgorgement techniques can be used to address a wide range of issues, such as reducing bias or toxicity, increasing fidelity, and ensuring responsible use of intellectual property. In this paper, we survey the landscape of model disgorgement methods and introduce a taxonomy of disgorgement techniques that are applicable to modern ML systems. In particular, we investigate the various meanings of “removing the effects” of data on the trained model in a way that does not require retraining from scratch.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

LanguageMachine Learning

Stichwörter der Autor:innen

artificial intelligencegenerative AImachine learningmachine unlearningmodel disgorgement

Autor:innen und Institutionen

  • Amazon Web Services Artificial Intelligence (AWS AI), Pasadena, CA 91125.
  • Computer and Information Science, University of Pennsylvania, Philadelphia, PA 19130.
  • Computer Science, University of California, Los Angeles, CA 90095.

ORCID-Profile: Carson Klingenberg

Zitieren

APA

Achille, A., Kearns, M., Klingenberg, C., & Soatto, S. (2024). AI model disgorgement: Methods and choices. Proceedings of the National Academy of Sciences, 121(18). https://doi.org/10.1073/pnas.2307304121

Vancouver

Achille A, Kearns M, Klingenberg C, Soatto S. AI model disgorgement: Methods and choices. Proceedings of the National Academy of Sciences. 2024;121(18). doi:10.1073/pnas.2307304121

BibTeX
@article{Achille2024,
  author = {Achille, Alessandro and Kearns, Michael and Klingenberg, Carson and Soatto, Stefano},
  title = {AI model disgorgement: Methods and choices},
  journal = {Proceedings of the National Academy of Sciences},
  year = {2024},
  volume = {121},
  number = {18},
  doi = {10.1073/pnas.2307304121}
}
Kostenlos starten →