Continual Learning With Participation Privacy: An Auditable Buffering-Aggregation Recipe

T-H. Hubert Chan, Elaine Shi, Mengshi Zhao, Mingxun Zhou
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:12550-12595, 2026.

Abstract

Modern federated and streaming learning systems often release intermediate models, so privacy must hold for the full trajectory under adaptive interaction. Motivated by participation privacy, we study single-edit neighboring user streams, where one insertion/deletion shifts all subsequent updates and defeats standard Hamming-neighbor continual-release analyses. We give an auditable modular recipe. A randomized buffering wrapper emits bins of size $[U,2U]$, reducing single-edit streams to a Hamming-style per-bin update stream with explicit backlog/delay guarantees, where $U$ is calibrated by the privacy parameters $(\varepsilon,\delta)$. We then prove a certification theorem for independently decomposable (prefix-causal, fresh-noise) continual mechanisms: any non-adaptive Hamming-neighbor DP proof lifts to adaptive inputs. Together, these ingredients yield trajectory-level $(\varepsilon,\delta)$-DP for single-edit streams using standard primitives (e.g., tree prefix sums), with an explicit privacy–latency link via $U$. Streaming DP-SGD experiments validate the privacy-utility-latency tradeoffs and the induced delay distributions.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-chan26b, title = {Continual Learning With Participation Privacy: An Auditable Buffering-Aggregation Recipe}, author = {Chan, T-H. Hubert and Shi, Elaine and Zhao, Mengshi and Zhou, Mingxun}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {12550--12595}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/chan26b/chan26b.pdf}, url = {https://proceedings.mlr.press/v306/chan26b.html}, abstract = {Modern federated and streaming learning systems often release intermediate models, so privacy must hold for the full trajectory under adaptive interaction. Motivated by participation privacy, we study single-edit neighboring user streams, where one insertion/deletion shifts all subsequent updates and defeats standard Hamming-neighbor continual-release analyses. We give an auditable modular recipe. A randomized buffering wrapper emits bins of size $[U,2U]$, reducing single-edit streams to a Hamming-style per-bin update stream with explicit backlog/delay guarantees, where $U$ is calibrated by the privacy parameters $(\varepsilon,\delta)$. We then prove a certification theorem for independently decomposable (prefix-causal, fresh-noise) continual mechanisms: any non-adaptive Hamming-neighbor DP proof lifts to adaptive inputs. Together, these ingredients yield trajectory-level $(\varepsilon,\delta)$-DP for single-edit streams using standard primitives (e.g., tree prefix sums), with an explicit privacy–latency link via $U$. Streaming DP-SGD experiments validate the privacy-utility-latency tradeoffs and the induced delay distributions.} }
Endnote
%0 Conference Paper %T Continual Learning With Participation Privacy: An Auditable Buffering-Aggregation Recipe %A T-H. Hubert Chan %A Elaine Shi %A Mengshi Zhao %A Mingxun Zhou %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-chan26b %I PMLR %P 12550--12595 %U https://proceedings.mlr.press/v306/chan26b.html %V 306 %X Modern federated and streaming learning systems often release intermediate models, so privacy must hold for the full trajectory under adaptive interaction. Motivated by participation privacy, we study single-edit neighboring user streams, where one insertion/deletion shifts all subsequent updates and defeats standard Hamming-neighbor continual-release analyses. We give an auditable modular recipe. A randomized buffering wrapper emits bins of size $[U,2U]$, reducing single-edit streams to a Hamming-style per-bin update stream with explicit backlog/delay guarantees, where $U$ is calibrated by the privacy parameters $(\varepsilon,\delta)$. We then prove a certification theorem for independently decomposable (prefix-causal, fresh-noise) continual mechanisms: any non-adaptive Hamming-neighbor DP proof lifts to adaptive inputs. Together, these ingredients yield trajectory-level $(\varepsilon,\delta)$-DP for single-edit streams using standard primitives (e.g., tree prefix sums), with an explicit privacy–latency link via $U$. Streaming DP-SGD experiments validate the privacy-utility-latency tradeoffs and the induced delay distributions.
APA
Chan, T.H., Shi, E., Zhao, M. & Zhou, M.. (2026). Continual Learning With Participation Privacy: An Auditable Buffering-Aggregation Recipe. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:12550-12595 Available from https://proceedings.mlr.press/v306/chan26b.html.

Related Material