A Recursive Hyperspheric Classification Algorithm

This paper presents a novel method for learning from a labeled dataset to accurately classify unknown data. The recursive algorithm, termed Recursive Hyperspheric Classification, or RHC, can accurately learn the classes of a labeled, n-dimensional dataset via a training method that recursively spawns a set of hyperspheres, endeavoring to separate and divide the feature space into partitions. This produces a comprehensive mapping of the space. These hyperspheres provide guidance for the search because they are recursively traversed. Some benchmarking has been performed on various data sets and has shown to yield superior results to more traditional artificial methods.

• Type: Conference
• Year: 2008
• Where: CAINE
• Authors: Salyer B. Reed, Carl G. Looney, Sergiu Dascalu
• Keywords: CAINE 2008, Computer Science, Hyperspheres Provide Guidance, N-dimensional Dataset, Recursive Hyperspheric Classification