Metraipolis › archive › dc_0b103bfdba0a

Evaluating Machine Unlearning in ASR — arXiv:2609.34092

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.34092
License (as archived)
CC0-1.0
Archived
2026-10-07T21:53:18.209Z
SHA-256 of the archived text
bb8d1f079c07be5212cb46c24c60bb4597ec44c5d27859c83e7a7e304d55dc4f
Stable URL of this page
https://metraipolis.com/documento/dc_0b103bfdba0a

How to cite

Evaluating Machine Unlearning in ASR — arXiv:2609.34092 — arXiv (en), https://arxiv.org/abs/2609.34092 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0b103bfdba0a (Metraipolis archive dc_0b103bfdba0a, sha256 bb8d1f079c07be5212cb46c24c60bb4597ec44c5d27859c83e7a7e304d55dc4f)
BibTeX
@misc{dc0b103bfdba0a,
  title = {Evaluating Machine Unlearning in ASR — arXiv:2609.34092},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34092},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0b103bfdba0a; sha256 bb8d1f079c07be5212cb46c24c60bb4597ec44c5d27859c83e7a7e304d55dc4f},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34092. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Machine unlearning (MU) offers a path to compliance with "right to be forgotten" regulations. While MU has received increasing attention for speech tasks, it remains largely unexplored for Automatic Speech Recognition (ASR). In this work, we investigate whether existing MU algorithms and evaluation tools are suitable for ASR. We apply several MU techniques to an ASR model, evaluating privacy-utility trade-offs for single-subject unlearning, then assess the best algorithm under sequential and simultaneous unlearning. Results show that gradient ascent-based algorithms achieve strong utility-privacy trade-offs, whereas more complex approaches over-unlearn samples, making them easier to identify as unlearned. This suggests standard privacy evaluations based on simple Membership Inference attacks are insufficient to reliably assess unlearning success, motivating improved evaluation methods for MU in ASR. Finally, we show that both sequential and simultaneous unlearning yield worse privacy and utility than single-subject unlearning, underscoring the need for unlearning constructions better suited to these settings.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.34092 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T21:27:08.584Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).