O2N2 framework enables doubly oblivious approximate nearest neighbor search
Original title: O2N2: Efficient Doubly Oblivious Approximate Nearest Neighbor Search
O2N2 achieves 9.8-125.9x speedup over prior doubly oblivious ANN baselines on real-world datasets.
Privacy-preserving ANN search on untrusted servers leaks access patterns through TEEs. O2N2 uses single-path search in rotated k-d tree forests with pointer-based Circuit ORAM for double obliviousness. It avoids trusted storage and scales to 100 million vectors with fixed-schedule operations for top-k results.
- Single-path search in randomly rotated k-d tree forests
- Separate storage for compact internal nodes and vector-bearing leaves
- 9.8-125.9x speedup for 10-NN and 100-NN at high-recall targets
- Scales to 100 million vectors
From the original article
1 min read at arxiv.orgPrivacy-preserving approximate nearest neighbor (ANN) search allows clients to search from an untrusted server without revealing sensitive queries, and is useful in scenarios like medical diagnostics and credit scoring.
By Zhechen Li · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.