Competence-guided Editing Methods for Lazy Learning

Lazy learning algorithms retain their raw training examples and defer all example-processing until problem solving time (eg, case-based learning, instance-based learning, and nearest-neighbour methods). A case-based classifier will typically compare a new target query to every case in its case-base (its raw training data) before deriving a target classification. This can make lazy methods prohibitively costly for large training sets. One way to reduce these costs is to filter or edit the original training set, to produce a reduced edited set by removing redundant or noisy examples. In this paper we describe and evaluate a new family of hybrid editing techniques that combine many of the features found in more traditional approaches with new techniques for estimating the usefulness of training examples. We demonstrate that these new techniques enjoy superior performance when compared to traditional and state-of-the-art methods.