SecFormer: Towards Fast and Accurate Privacy-Preserving Inference for Large Language Models.
Jinglong LuoYehong ZhangJiaqi ZhangXin MuHui WangYue YuZenglin XuPublished in: CoRR (2024)
Keyphrases
- privacy preserving
- language model
- language modeling
- privacy preserving data mining
- document retrieval
- privacy preservation
- vertically partitioned data
- n gram
- probabilistic model
- language modelling
- query expansion
- retrieval model
- information retrieval
- multi party
- statistical language models
- test collection
- smoothing methods
- data privacy
- privacy concerns
- sensitive information
- privacy sensitive
- horizontally partitioned data
- private information
- query terms
- partitioned data
- private data
- privacy protection
- privacy preserving association rule mining
- relevance model
- privacy requirements
- secure multiparty computation
- bayesian networks
- sensitive attributes
- preserving privacy
- scalar product