Incremental learning from examples using HC-expressions

摘要

An incremental learning algorithm for learning from examples using the data structure Hierarchical Counterfactual (HC) expression is presented. The data structure used, Hierarchical Counterfactual expressions, is tree like and can be easily modified. Alternate levels in the tree are positive and negative exceptions. The multilevel exceptions make it ideal for incremental learning and can handle disjunctive and conjunctive concepts efficiently. The concept descriptions learnt are simple for both disjunctive and conjunctive concepts and express any existing internal structure in the data.