mlsanityData to 7 Oct 2026

Fully Decentralized Certified Unlearning

arXiv 2512.08443 ↗First posted 9 Dec 2025cs.LGcited by 1

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

Machine unlearning (MU) seeks to remove the influence of specified data from a trained model in response to privacy requests or data poisoning. While certified unlearning has been analyzed in centralized and server-orchestrated federated settings (via guarantees analogous to differential privacy, DP), the decentralized setting -- where peers communicate without a coordinator remains underexplored. We study certified unlearning in decentralized networks with fixed topologies and propose RR-DU, a random-walk procedure that performs one projected gradient ascent step on the forget set at the unlearning client and a geometrically distributed number of projected descent steps on the retained data elsewhere, combined with subsampled Gaussian noise and projection onto a trust region around the original model. We provide (i) convergence guarantees in the convex case and stationarity guarantees in the nonconvex case, (ii) (varepsilon,δ) network-unlearning certificates on client views via subsampled Gaussian R'enyi DP (RDP) with segment-level subsampling, and (iii) deletion-capacity bounds that scale with the forget-to-local data ratio and quantify the effect of decentralization (network mixing and randomized subsampling) on the privacy-utility trade-off. Empirically, on image benchmarks (MNIST, CIFAR-10), RR-DU matches a given (varepsilon,δ) while achieving higher test accuracy than decentralized DP baselines and reducing forget accuracy to random guessing (≈ 10%).

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 32 of its 52 references that are in the map. First posted on arXiv 9 Dec 2025.

Builds on

32 of 52 references.

Closest ancestors · 9

Same direction, newest first

  1. 2025Certified Unlearning for Neural Networks8 Jun
  2. 2025When to Forget? Complexity Trade-offs in Machine Unlearning24 Feb
  3. 2024The Utility and Complexity of in- and out-of-Distribution Machine Unlearning12 Dec
  4. 2024Rewind-to-Delete: Certified Machine Unlearning for Nonconvex Functions15 Sep
  5. 2024Towards Certified Unlearning for Deep Neural Networks1 Aug
  6. 2024Langevin Unlearning: A New Perspective of Noisy Gradient Descent for Machine Unlearning18 Jan
  7. 2023Tight Bounds for Machine Unlearning via Differential Privacy2 Sep
  8. 2022Forget Unlearning: Towards True Data-Deletion in Machine Learning17 Oct
  9. 2021Remember What You Want to Forget: Algorithms for Machine Unlearning4 Mar

Other references · 23

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2016Communication-Efficient Learning of Deep Networks from Decentralized Datacited by 7,592
  3. 2016Deep Learning with Differential Privacycited by 2,682
  4. 2017BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chaincited by 1,052
  5. 2020Inverting Gradients -- How easy is it to break privacy in federated learning?cited by 707
  6. 2019Certified Data Removal from Machine Learning Modelscited by 512
  7. 2019Making AI Forget You: Data Deletion in Machine Learningcited by 447
  8. 2020A Unified Theory of Decentralized SGD with Changing Topology and Local Updatescited by 361
  9. 2018Privacy Amplification by Subsampling: Tight Analyses via Couplings and Divergencescited by 237
  10. 2018Amplification by Shuffling: From Local to Central Differential Privacy via Anonymitycited by 163
  11. 2017Asynchronous Decentralized Parallel Stochastic Gradient Descentcited by 159
  12. 2018Privacy Amplification by Iterationcited by 97
11 more
  1. 2022Federated Unlearning: How to Efficiently Erase a Client in FL?cited by 90
  2. 2019The Privacy Blanket of the Shuffle Modelcited by 73
  3. 2023A Survey on Federated Unlearning: Challenges, Methods, and Future Directionscited by 58
  4. 2020Decentralized gradient methods: does topology matter?cited by 35
  5. 2016Gossip Dual Averaging for Decentralized Optimization of Pairwise Functionscited by 21
  6. 2024The Privacy Power of Correlated Noise in Decentralized Learningcited by 17
  7. 2024Communication Efficient and Provable Federated Unlearningcited by 16
  8. 2023Get Rid Of Your Trail: Remotely Erasing Backdoors in Federated Learningcited by 12
  9. 2019Differentially Private Summation with Multi-Message Shufflingcited by 9
  10. 2025Privacy Amplification Through Synthetic Data: Insights from Linear Regressioncited by 2
  11. 2025Unified Privacy Guarantees for Decentralized Learning via Matrix Factorizationcited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Priority-Aware Learning-Unlearning Correction for Dynamic Decentralized LoRA Fine-Tuningcited by 0