arXiv cs.AI· Peichun Hua, Danyang Chen, Junan Zhang, Haifeng Sun, Jingyu Wang, Diwen Xue, Mingyu Li, Yunming Xiao·· 3 小时前AI 评分47
Pointing the Way, Hiding the Destination：大规模实用私有稠密检索方案
Pointing the Way, Hiding the Destination: Practical Private Dense Retrieval at Scale
AI 导读
arXiv 论文提出将学习型深度哈希用作隐私过滤器，实现既保护查询与结果隐私、又只返回授权文档的私有稠密检索。用 200-500 个候选即可在 25K 至 5.4M 文档的五个零样本语料上接近全库检索质量；在 2.68M 段落的 NQ 语料、10-Gbps 链路上，仅给 128-token Qwen3-32B RAG 流水线增加 0.38 秒（6.5%）开销。
正文 · 原文
Abstract:Hosted retrieval-augmented generation (RAG) and semantic search allow users to query valuable provider-held corpora, raising two competing demands: to hide each query and chosen result, yet reveal only the documents that the user is authorized to receive. Existing cryptographic approaches either make this costly by processing the entire corpus for every query, or sacrifice quality for efficiency by scanning a few clusters. We repurpose learned deep hashing as a private filter: a randomized binary code points the provider to a short candidate list, while encrypted reranking and oblivious key transfer protect the precise query and final selection. This shortlist short-circuits full-corpus cryptographic search without sacrificing retrieval quality: with 200-500 candidates, it closely matches full-corpus retrieval across five zero-shot corpora spanning 25K to 5.4M documents. On the full 2.68M-passage NQ corpus over a 10-Gbps link, our protocol only adds 0.38 seconds, or 6.5%, to a 128-token Qwen3-32B RAG pipeline. The released code satisfies directional metric differential privacy (DP) and substantially reduces embedding-inversion and property-inference leakage, demonstrating that a carefully learned shortlist can make private dense retrieval both accurate and practical.
| Comments: | 33 pages, 9 figures, 18 tables |
| Subjects: | Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Machine Learning (cs.LG) |
| Cite as: | arXiv:2608.25735 [cs.CR] |
| (or arXiv:2608.25735v3 [cs.CR] for this version) | |
| https://doi.org/10.48550/arXiv.2608.25735 arXiv-issued DOI via DataCite |
Submission history
From: Peichun Hua [view email]
[v1]
Wed, 26 Aug 2026 12:46:36 UTC (419 KB)
[v2]
Sat, 3 Oct 2026 15:53:35 UTC (419 KB)
[v3]
Thu, 12 Nov 2026 13:59:25 UTC (425 KB)
来源：arXiv cs.AI · arxiv.org