Metraipolis › archive › dc_093aa133eff8

Exact Unlearning via Quantized Sufficient Statistics — arXiv:2610.07197

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.07197
License (as archived)
CC0-1.0
Archived
2026-10-07T11:40:54.663Z
SHA-256 of the archived text
607f5cc4bb87c38aedef8375cff76439bbaf5927a83e058c29c8d41690f5bbdc
Stable URL of this page
https://metraipolis.com/documento/dc_093aa133eff8

How to cite

Exact Unlearning via Quantized Sufficient Statistics — arXiv:2610.07197 — arXiv (en), https://arxiv.org/abs/2610.07197 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_093aa133eff8 (Metraipolis archive dc_093aa133eff8, sha256 607f5cc4bb87c38aedef8375cff76439bbaf5927a83e058c29c8d41690f5bbdc)
BibTeX
@misc{dc093aa133eff8,
  title = {Exact Unlearning via Quantized Sufficient Statistics — arXiv:2610.07197},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.07197},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_093aa133eff8; sha256 607f5cc4bb87c38aedef8375cff76439bbaf5927a83e058c29c8d41690f5bbdc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.07197. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Exact unlearning requires a deployed predictor to match one rebuilt without the information named by a deletion request. Existing general-purpose exact methods localize retraining through disjoint shards, but every request still invalidates a model, and smaller shards reduce the data available to each constituent predictor. We introduce Quantized Sufficient Statistics (QSS), which separates a small frozen schema from mutable, sum-decomposable content. The schema learns global structure; the content stores local prediction corrections as additive statistics indexed by quantized regions. Deleting content is therefore exact subtraction rather than optimization. We distinguish two guarantees: QSS-L exactly removes a label while retaining the unlabelled input, whereas QSS-E exactly removes both input and label by learning the schema without deletable examples. A deletion takes the arithmetic fast path with probability $1-\rho$ and triggers a full rebuild with probability $\rho$; all reported expected latencies include both events. Across 15 vision, text, and tabular datasets at $\rho=0.5\%$, QSS-L is within 2 percentage points of SISA on 11 tasks and provides 4--483$\times$ lower expected deletion latency on the low-class-count tasks where a compact schema is effective. QSS-E quantifies the additional accuracy cost of removing every trace of an input.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.07197 — 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 2026-10-07T11:19:24.734Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).