mlsanityData to 14 Oct 2026

Spectrum-Aware Bounds on Invertibility for Privacy-Enhancing Instance Encoding

arXiv 2608.23382 ↗First posted 24 Aug 2026cs.LG cs.CRcited by 0

Not on the map. Why?

Abstract (arXiv)

Instance encoding is a popular empirical technique for privacy enhancement when sharing data to an untrusted server. It transforms sensitive data through an encoding process before sharing, with the hope that the encoding process retains utility but makes it hard to reconstruct the original data. However, most work offers no theoretical guarantee that the encoding process is actually irreversible. A recent work derived a mean-squared error (MSE) bound limiting any adversary's reconstruction accuracy, offering one of the first theoretical results in this domain. This bound, however, has three critical limitations: it is often too loose, only works with randomized encoders (excluding many deterministic encoders practitioners use), and only bounds MSE. We introduce a family of new bounds that (1) are tighter, (2) applicable even to fully deterministic encoders, and (3) can extend beyond MSE to other norm-based similarity metrics, by properly accounting for the encoder's spectral structure. We evaluate our bounds across a range of encoders, datasets, and attacks, showing they hold consistently and improve upon the existing bound.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 37 of its 64 references that are in the map. First posted on arXiv 24 Aug 2026.

Builds on

37 of 64 references.

References · 37

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  3. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
  4. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  5. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 9,122
  6. 2019Certified Adversarial Robustness via Randomized Smoothingcited by 1,340
  7. 2017Deep Image Priorcited by 1,098
  8. 2014NICE: Non-linear Independent Components Estimationcited by 1,058
  9. 2014Understanding Deep Image Representations by Inverting Themcited by 403
  10. 2018Split learning for health: Distributed deep learning without sharing raw patient datacited by 395
  11. 2019Sliced Score Matching: A Scalable Approach to Density and Score Estimationcited by 384
  12. 2020SplitFed: When Federated Learning Meets Split Learningcited by 310
25 more
  1. 2018Towards Fast Computation of Certified Robustness for ReLU Networkscited by 255
  2. 2020InstaHide: Instance-hiding Schemes for Private Distributed Learningcited by 90
  3. 2019Split Learning for collaborative deep learning in healthcarecited by 73
  4. 2023DP-Forward: Fine-tuning and Inference on Language Models with Differential Privacy in Forward Passcited by 67
  5. 2020NoPeek: Information leakage reduction to share activations in distributed deep learningcited by 66
  6. 2022Bounding Training Data Reconstruction in Private (Deep) Learningcited by 46
  7. 2022ResSFL: A Resistance Transfer Framework for Defending Model Inversion Attack in Split Federated Learningcited by 36
  8. 2021Measuring Data Leakage in Machine-Learning Models with Fisher Informationcited by 33
  9. 2023Split-and-Denoise: Protect large language model inference with local differential privacycited by 32
  10. 2018Deep Private-Feature Extractioncited by 30
  11. 2021Practical Defences Against Model Inversion Attacks for Split Neural Networkscited by 29
  12. 2018Not Just Privacy: Improving Performance of Private Deep Learning in Mobile Cloudcited by 28
  13. 2020DISCO: Dynamic and Invariant Sensitive Channel Obfuscation for deep neural networkscited by 23
  14. 2022PAC Privacy: Automatic Privacy Measurement and Control of Data Processingcited by 19
  15. 2019DeepObfuscator: Obfuscating Intermediate Representations with Privacy-Preserving Adversarial Learning on Smartphonescited by 15
  16. 2022SplitNets: Designing Neural Architectures for Efficient Distributed Computing on Head-Mounted Systemscited by 14
  17. 2023Bounding the Invertibility of Privacy-preserving Instance Encoding using Fisher Informationcited by 13
  18. 2016Inferential Privacy Guarantees for Differentially Private Mechanismscited by 11
  19. 2021NeuraCrypt: Hiding Private Health Data via Random Neural Networks for Public Trainingcited by 11
  20. 2019Interpretable Complex-Valued Neural Networks for Privacy Protectioncited by 10
  21. 2021NeuraCrypt is not privatecited by 10
  22. 2023PATROL: Privacy-Oriented Pruning for Collaborative Inference Against Model Inversion Attackscited by 9
  23. 2022Syfer: Neural Obfuscation for Private Data Releasecited by 3
  24. 2021Enabling Inference Privacy with Adaptive Noise Injectioncited by 1
  25. 2023Intrinsic Bayesian Cram'er-Rao Bound with an Application to Covariance Matrix Estimationcited by 1

Built on by

Not cited yet.