Faster Vector Searches with Approximate Nearest Neighbor index -- now available in the Vector Search Early Access Program

We've recently made available a new version of InterSystems IRIS in the Vector Search Early Access Program, featuring a new Approximate Nearest Neighbor index based upon the Hierarchical Navigable Small World (HNSW) indexing algorithm. This addition allows for highly efficient, approximate nearest-neighbor searches over large vector datasets, dramatically improving query performance and scalability.

The HNSW algorithm is designed to optimize vector search for high-dimensional data by building a graph-based structure, making it faster to find approximate neighbors in large collections of vectors. Whether you’re working with recommendation systems, natural language processing, or other machine learning applications, HNSW can significantly reduce search times while allowing you to tune the level of accuracy with the tradeoff that higher accuracy results in slower query times.

Key benefits of HNSW include:

    •    Faster searches even as dataset size grows
    •    Reduced memory footprint while maintaining high accuracy
    •    Seamless integration with existing IRIS vector search capabilities

How to Get Started

The latest version is now available through our Vector Search Early Access Program. To participate, sign up here, download the new version, and begin testing. Your feedback is critical as we continue to enhance Vector Search!

We encourage you to explore the performance improvements and share your thoughts with the community. Please contact me with any questions or feedback you have during the early access phase.

Happy coding!