发表机构
Arizona State University; McGill University（亚利桑那州立大学; 麦吉尔大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对机器遗忘中数学保证与部署工件脱节的问题，提出ExecCert发布时认证层，通过原生证书或RRV验证候选工件，实现增量认证并降低成本，实验证明其有效性。
AI 中文摘要
当数据因删除请求、过时记录或数据质量问题而必须移除时，需要机器遗忘，而从头重新训练可能代价高昂。经过认证的机器遗忘方法提供数学保证，而部署的系统会发布由软件产生的具体有限精度工件。为弥合数学保证与实际部署之间的差距，我们引入了可执行发布认证（ExecCert），这是一种发布时层，用于认证考虑发布的候选工件。ExecCert 要么为已执行的候选关闭方法自身的原生证书，要么应用重训练参考发布验证（RRV）来认证对当前保留集重训练的保真度。顺序删除使后者变得不平凡，因为精确的保留集参考和存储的数值状态分别演化。对于具有可变岭头的冻结表示，我们开发了 RRV 的增量实现，该实现跨删除请求维护认证证据，而不是在每次发布时重建它。在四个已发布的遗忘实现上，ExecCert 保留了有效证书，更改了发布决策，收紧了保守界限，并识别了具体输出所支持的重训练参考保真度。在顺序服务实验中，RRV 消除了由存储方程验证引起的错误发布，同时紧密跟踪实际误差，并且一旦发布检查变得足够频繁，增量认证比新鲜和维护的验证因子替代方案都更便宜。
英文摘要
Machine unlearning is needed when data must be removed because of deletion requests, outdated records, or data-quality concerns, while retraining from scratch can be costly. Certified machine unlearning methods provide mathematical guarantees, while deployed systems release concrete finite-precision artifacts produced by software. To bridge the gap between mathematical guarantees and practical deployment, we introduce Executable Release Certification (ExecCert), a release-time layer that certifies the candidate artifact considered for release. ExecCert either closes a method's native certificate for the executed candidate or applies Retraining-Reference Release Verification (RRV) to certify fidelity to current retain-set retraining. Sequential deletion makes the latter nontrivial because the exact retain-set reference and the stored numerical state evolve separately. For frozen representations with a mutable ridge head, we develop an incremental realization of RRV that maintains certified evidence across deletion requests rather than reconstructing it at each release. On four published unlearning implementations, ExecCert preserves valid certificates, changes release decisions, tightens conservative bounds, and identifies the retraining-reference fidelity supported by concrete outputs. In sequential-service experiments, RRV eliminates false releases caused by stored-equation verification while closely tracking realized error, and incremental certification remains cheaper than both fresh and maintained verified-factor alternatives once release checks become sufficiently frequent.
