Distance Based
Distance-based methods are increasingly used in machine learning for classification, clustering, and other tasks, focusing on quantifying the similarity or dissimilarity between data points or distributions. Current research explores novel distance metrics tailored to specific data types (e.g., distributions, time series, categorical variables) and integrates them into various algorithms like k-nearest neighbors and clustering techniques, often alongside dimensionality reduction methods to improve efficiency and accuracy. These advancements enhance the interpretability and computational efficiency of machine learning models across diverse fields, including astronomy, forensics, and robotics, while also providing new tools for analyzing complex data structures like social networks and biological images.
Papers
Hubs and Hyperspheres: Reducing Hubness and Improving Transductive Few-shot Learning with Hyperspherical Embeddings
Daniel J. Trosten, Rwiddhi Chakraborty, Sigurd Løkse, Kristoffer Knutsen Wickstrøm, Robert Jenssen, Michael C. Kampffmeyer
Evaluation of distance-based approaches for forensic comparison: Application to hand odor evidence
Isabelle Rivals, Cédric Sautier, Guillaume Cognon, Vincent Cuzuel