发表机构
Meta Platforms, Inc.（Meta平台公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出量化充分统计量（QSS），通过分离冻结模式与可加性统计量实现精确遗忘，删除为精确减法，在15个数据集上显著降低延迟并保持准确性。
AI 中文摘要
精确遗忘要求部署的预测器与在删除请求所指定的信息未参与的情况下重建的预测器相匹配。现有的通用精确方法通过不相交的分片来局部化重训练，但每次请求仍会使模型失效，且更小的分片会减少每个组成预测器可用的数据。我们引入了量化充分统计量（QSS），它将一个小的冻结模式与可变的、可求和分解的内容分离。模式学习全局结构；内容将局部预测修正存储为按量化区域索引的加性统计量。因此，删除内容就是精确的减法而非优化。我们区分两种保证：QSS-L精确移除标签而保留未标记输入，而QSS-E通过在不包含可删除示例的情况下学习模式来精确移除输入和标签。一次删除以概率$1-\ ho$走算术快速路径，以概率$\ ho$触发完全重建；所有报告的期望延迟都包含这两种事件。在15个视觉、文本和表格数据集上，当$\ ho=0.5\%$时，QSS-L在11个任务上比SISA低2个百分点以内，并在低类别数任务上提供4--483倍的更低期望删除延迟，在这些任务中紧凑模式有效。QSS-E量化了移除输入所有痕迹的额外准确性成本。
英文摘要
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 $1-ρ$ and triggers a full rebuild with probability $ρ$; all reported expected latencies include both events. Across 15 vision, text, and tabular datasets at $ρ=0.5\%$, 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.
Comments33 pages, 20 figures, 15 tables. Accepted at NeurIPS 2026
