mlsanityData to 7 Oct 2026

Split-and-Denoise: Protect large language model inference with local differential privacy

arXiv 2310.09130 ↗First posted 13 Oct 2023cs.AI cs.CRcited by 30

Not on the map. Why?

Abstract (arXiv)

Large Language Models (LLMs) excel in natural language understanding by capturing hidden semantics in vector space. This process enriches the value of text embeddings for various downstream tasks, thereby fostering the Embedding-as-a-Service (EaaS) business model. However, the risk of privacy leakage due to direct text transmission to servers remains a critical concern. To address this, we introduce Split-N-Denoise (SnD), an private inference framework that splits the model to execute the token embedding layer on the client side at minimal computational cost. This allows the client to introduce noise prior to transmitting the embeddings to the server, and subsequently receive and denoise the perturbed output embeddings for downstream tasks. Our approach is designed for the inference stage of LLMs and requires no modifications to the model parameters. Extensive experiments demonstrate SnD's effectiveness in optimizing the privacy-utility tradeoff across various LLM architectures and diverse downstream tasks. The results reveal an improvement in performance under the same privacy budget compared to the baselines by over 10% on average, offering clients a privacy-preserving solution for local privacy protection.

Key facts. Cited by 30 papers in the map (data through 7 Oct 2026). Builds on 39 of its 85 references that are in the map. First posted on arXiv 13 Oct 2023.

Builds on

39 of 85 references.

References · 39

  1. 2017Attention Is All You Needcited by 59,932
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  3. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
  4. 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,243
  5. 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,115
  6. 2016Pointer Sentinel Mixture Modelscited by 2,992
  7. 2015Character-level Convolutional Networks for Text Classificationcited by 2,293
  8. 2018Neural Network Acceptability Judgmentscited by 952
  9. 2017DailyDialog: A Manually Labelled Multi-turn Dialogue Datasetcited by 649
  10. 2020Efficient Intent Detection with Dual Sentence Encoderscited by 419
  11. 2020TweetEval: Unified Benchmark and Comparative Evaluation for Tweet Classificationcited by 414
  12. 2021Differentially Private Fine-tuning of Language Modelscited by 392
27 more
  1. 2018Split learning for health: Distributed deep learning without sharing raw patient datacited by 391
  2. 2013Good Debt or Bad Debt: Detecting Semantic Orientations in Economic Textscited by 310
  3. 2023Analyzing Leakage of Personally Identifiable Information in Language Modelscited by 278
  4. 2018Distributed learning of deep neural network over multiple agentscited by 277
  5. 2018Improving the Gaussian Mechanism for Differential Privacy: Analytical Calibration and Optimal Denoisingcited by 216
  6. 2020Explainable Automated Fact-Checking for Public Health Claimscited by 209
  7. 2018Hate Speech Dataset from a White Supremacy Forumcited by 196
  8. 2019Privacy- and Utility-Preserving Textual Analysis via Calibrated Multivariate Perturbationscited by 115
  9. 2023Flocks of Stochastic Parrots: Differentially Private Prompt Learning for Large Language Modelscited by 80
  10. 2019Detailed comparison of communication efficiency of split learning and federated learningcited by 76
  11. 2023DP-Forward: Fine-tuning and Inference on Language Models with Differential Privacy in Forward Passcited by 65
  12. 2022THE-X: Privacy-Preserving Transformer Inference with Homomorphic Encryptioncited by 55
  13. 2021Natural Language Understanding with Privacy-Preserving BERTcited by 54
  14. 2018Generalised Differential Privacy for Text Document Processingcited by 52
  15. 2020Investigating Societal Biases in a Poetry Composition Systemcited by 47
  16. 2020Differentially Private Language Models Benefit from Public Pre-trainingcited by 46
  17. 2022Bounding Training Data Reconstruction in Private (Deep) Learningcited by 43
  18. 2023Privacy-Preserving Parameter-Efficient Fine-Tuning for Large Language Model Servicescited by 43
  19. 2023Hide and Seek (HaS): A Lightweight Framework for Prompt Privacy Protectioncited by 37
  20. 2020TextHide: Tackling Data Privacy in Language Understanding Taskscited by 32
  21. 2020Improving Deep Learning with Differential Privacy using Gradient Encoding and Denoisingcited by 30
  22. 2022Differentially Private Label Protection in Split Learningcited by 25
  23. 2023Protecting User Privacy in Remote Conversational Systems: A Privacy-Preserving framework based on text sanitizationcited by 23
  24. 2023LLMs Can Understand Encrypted Prompt: Towards Privacy-Computing Friendly Transformerscited by 20
  25. 2019DP-LSSGD: A Stochastic Optimization Method to Lift the Utility in Privacy-Preserving ERMcited by 15
  26. 2021Representations of Language Varieties Are Reliable Given Corpus Similarity Measurescited by 10
  27. 2022An Efficient Industrial Federated Learning Framework for AIoT: A Face Recognition Applicationcited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 1; 2024: 7; 2025: 7; 2026: 15.

30 papers.

Sort:

By area:

  1. 2024Large Language Model-Brained GUI Agents: A Surveycited by 160
  2. 2024Navigating LLM Ethics: Advancements, Challenges, and Future Directionscited by 28
  3. 2024Deconstructing The Ethics of Large Language Models from Long-standing Issues to New-emerging Dilemmas: A Surveycited by 23
  4. 2025Position: Privacy Is Not Just Memorization!cited by 15
  5. 2024Provable Differentially Private Computation of the Cross-Attention Mechanismcited by 14
  6. 2024De-mark: Watermark Removal in Large Language Modelscited by 11
  7. 2024Personalized Federated Fine-Tuning for LLMs via Data-Driven Heterogeneous Model Architecturescited by 10
  8. 2024Delving into Differentially Private Transformercited by 5
  9. 2025A Novel Hat-Shaped Device-Cloud Collaborative Inference Framework for Large Language Modelscited by 5
  10. 2023ConfusionPrompt: Practical Private Inference for Online Large Language Modelscited by 3
  11. 2025MaskSQL: Safeguarding Privacy for LLM-Based Text-to-SQL via Abstractioncited by 3
  12. 2025Learning Obfuscations Of LLM Embedding Sequences: Stained Glass Transformcited by 2
18 more
  1. 2026Towards Privacy-Preserving LLM Inference via Covariant Obfuscation (Technical Report)cited by 2
  2. 2025LLM in the Middle: A Systematic Review of Threats and Mitigations to Real-World LLM-based Systemscited by 1
  3. 2025Federated Attention: A Distributed Paradigm for Collaborative LLM Inference over Edge Networkscited by 1
  4. 2026SoK: Privacy-aware LLM in Healthcare: Threat Model, Privacy Techniques, Challenges and Recommendationscited by 1
  5. 2026NOIR: Privacy-Preserving Generation of Code with Open-Source LLMscited by 1
  6. 2026Towards Privacy-Preserving Machine Translation at the Inference Stage: A New Task and Benchmarkcited by 1
  7. 2026PAAC: Privacy-Aware Agentic Device-Cloud Collaborationcited by 1
  8. 2026From Prompts to Responses: Dual-Sided Data Leakage and Defense in Split Large Language Modelscited by 1
  9. 2026The Art of Mixology: Mixup-based Obfuscation for Privacy-Preserving Split Learning in Large Language Modelscited by 1
  10. 2025Stereo-LiDAR Fusion by Semi-Global Matching With Discrete Disparity-Matching Cost and Semidensificationcited by 0
  11. 2026Differentially Private and Communication Efficient Large Language Model Split Inference via Stochastic Quantization and Soft Promptcited by 0
  12. 2026Privacy-Aware Split Inference with Speculative Decoding for Large Language Models over Wide-Area Networkscited by 0
  13. 2026Neural Gate: Mitigating Privacy Risks in LVLMs via Neuron-Level Gradient Gatingcited by 0
  14. 2026Privacy from Symmetry: Orthogonally Equivariant Transformers for LLM Inferencecited by 0
  15. 2026Efficient and Privacy Aware Edge Cloud Collaborative Inference for Large Language Modelscited by 0
  16. 2026P2Skill: Privacy Preserving Skill Distillation for Cloud-Local LLM Inference Systemscited by 0
  17. 2026Spectrum-Aware Bounds on Invertibility for Privacy-Enhancing Instance Encodingcited by 0
  18. 2026Learning to Refer: Client-Resolved Generation for Privacy-Aware Language Modelscited by 0