approximate-nearest-neighbor

VectorRAG.Net is a .NET-native high-performance vector database library for semantic search and RAG (Retrieval-Augmented Generation). Core search is based on Random Hyperplane LSH candidate generation with exact rerank by dot/cosine.

Training-free ordinal & sign quantization for compressed nearest-neighbour retrieval over high-dimensional embeddings. Pure Rust, zero system dependencies.

Near-optimal vector quantization from Google's ICLR 2026 paper — 95% recall, 5x compression, zero preprocessing, pure Python FAISS replacement

PostgreSQL TurboQuant Index for PGVector

High-performance database management system

Open vector indexes for the SQL engines you already use.

⚡ Blazing-fast GPU vector search for PostgreSQL powered by NVIDIA cuVS.

DiskANN vs HNSW: a measured vector-index benchmark at 3.38M and 10K scale (recall/QPS/memory, concurrency, reproducible harness)

Hybrid B+ Tree and HNSW index for efficient k-NN search with scalar filtering using probabilistic optimization

Comparing attribute-set representations (linear/blocked bitsets, Roaring) for filtered HNSW search.

TurboQuant-style HNSW benchmark on GloVe-200: Recall@10 vs compression across 1-8 bits.

AI-powered vector database with HNSW indexing, FastAPI server, WAL persistence, distributed sharding, and RAG pipeline integration

Python library for MUVERA multi-vector retrieval via Fixed Dimensional Encodings. ColBERT / ColQwen2 / ColQwen3.5 compatible.

High-performance pure Go HNSW vector search with persistence, filtering, and segment merging

Protein homology search using transformer-based embeddings and Approximate Nearest Neighbor methods for efficient biological similarity detection

All nearest-neighbor search using neighborhood graphs

Self-supervised maritime anomaly detection using Transformer representation learning, HNSW behavioral retrieval, and multimodal AIS trajectory intelligence.

Approximate Nearest Neighbor search implementation in C, including LSH, Hypercube, IVF, and Product Quantization algorithms

Learning-based Approximate Nearest Neighbor search using neural classifiers for partition prediction and efficient retrieval