Privacy-Preserving LLM Embedding Transmission for End-Cloud Collaboration
cs.CR
Submitted: 2025-03-17
Updated: 2026-08-30
Terminology
Sources
- The Llama 3 Herd of Models
- Decepticons: Corrupted Transformers Breach Privacy in Federated Learning for Language Models
- Explaining and Harnessing Adversarial Examples
- Adam: A Method for Stochastic Optimization
- Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generation
- Sentence Embedding Leaks More Information than You Expect: Generative Embedding Inversion Attack to Recover the Whole Sentence
- Privacy-Preserving Parameter-Efficient Fine-Tuning for Large Language Model Services
- Eguard: Defending LLM Embeddings Against Inversion Attacks via Text Mutual Information Optimization
- RoBERTa: A Robustly Optimized BERT Pretraining Approach
- Towards Deep Learning Models Resistant to Adversarial Attacks
- Efficient Estimation of Word Representations in Vector Space
- Text Embeddings Reveal (Almost) As Much As Text
- Language Model Inversion
- Large Dual Encoders Are Generalizable Retrievers
- BEIR: A Heterogenous Benchmark for Zero-shot Evaluation of Information Retrieval Models
- FEVER: a large-scale dataset for Fact Extraction and VERification
- LLaMA: Open and Efficient Foundation Language Models
- Information Leakage from Embedding in Large Language Models
- Crafter: Facial Feature Crafting against Inversion-based Identity Theft on Deep Models
- A Differentially Private Text Perturbation Method Using a Regularized Mahalanobis Metric
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs