Fuzzy art variants for improved single pass sequential clustering of data
G.P. Noone · 2006
We often have some strong a priori information regarding the expected variations of measured data. For example, radar pulse parameters as measured by modern Electronic Support Receivers [3]. We use this information to our advantage when building Fuzzy ART variants to sequentially sort such parameters from each radar into distinct categories. One of the variants maintains the full Fuzzy ART structure and all inputs and outputs are constrained to be between 0 and 1. The second variant has no such restrictions but its computation is still based mainly on elementary, fuzzy operations. We show that our variants considerably improve the clustering of simulated 2-D data. The variants are very general in nature and can be used for any single pass data clustering problem in which we have some a priori information of the variation of the parameters that constitute the data.