Learning structural descriptions of patterns: A new technique for conditional clustering and rule generation

摘要

A deterministic technique is developed for generating rules which can optimally classify patterns (for example, in 3D object recognition) in terms of the bounds on unary (single part) and binary (part relation) features which constitute different types of patterns. This technique, termed Conditional Rule Generation (CRG), was developed to take into account the label-compatibilities which should occur between unary and binary rules in their very generation, a condition which is generally not guaranteed in well-known rule generation and machine learning techniques.