Holo-NN: New algebraic framework for private nearest neighbor search

Researchers have introduced Holo-NN, a novel algebraic framework for privacy-preserving approximate nearest neighbor (ANN) search in high-dimensional embeddings. This method replaces traditional branching operations, which are difficult to implement with fully homomorphic encryption, with a single linear projection. By combining compressive sensing with locality-preserving feature transforms, Holo-NN folds databases into a compact holographic sensing matrix, enabling encrypted searches through a single matrix-vector multiplication. This approach avoids complex cryptographic bootstrapping and multi-round interactions, making private retrieval an efficient, parallelizable computation suitable for both CPUs and GPUs. AI

IMPACT Introduces a novel privacy-preserving search method that could enhance security in AI applications dealing with sensitive data.

RANK_REASON The cluster contains a research paper detailing a new method for nearest neighbor search.

Read on arXiv cs.IR (Information Retrieval) →

• alphaXiv
• arXiv
• CatalyzeX
• compressed sensing
• DagsHub
• fully homomorphic encryption
• Gotit.pub
• Hierarchical Navigable Small World graphs
• Holo-NN
• Hugging Face
• Scalable Nearest Neighbors
• ScienceCast
• CORE Recommender

• paper
• safety

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →