Nearest Neighbor Graph

Nearest neighbor graphs (NNGs) represent data points by connecting each point to its closest neighbors, forming a network that captures local data structure. Current research focuses on optimizing NNG construction, including adaptive k-value selection, pruning spurious edges to improve manifold learning, and leveraging hypergraph extensions to model higher-order relationships. These advancements enhance the utility of NNGs in diverse applications such as dimensionality reduction, data interpolation, clustering, and information extraction from complex documents, improving accuracy and efficiency in these tasks.

Papers