Rule Induction for Classification Using Multi-objective Genetic Programming

Abstract

Abstract. Multi-objective metaheuristics have previously been applied to partial classification, where the objective is to produce simple, easy to understand rules that describe subsets of a class of interest. While this provides a useful aid in descriptive data mining, it is difficult to see how the rules produced can be combined usefully to make a predictive classifier. This paper describes how, by using a more complex representation of the rules, it is possible to produce effective classifiers for two class problems. Furthermore, through the use of multi-objective genetic programming, the user can be provided with a selection of classifiers providing different trade-offs between the misclassification costs and the overall model complexity.

MOCA-I: Discovering Rules and Guiding Decision Maker in the Context of Partial Classification in Large and Imbalanced Datasets

“…Given two equivalent rules, the simplest must be preferred. The addition of one objective promoting simpler rules is a common solution, successfully applied in Reynolds and Iglesia work [10]. In addition to this, Barcadit et al used rule deletion operators [11].…”

Cultural Algorithm Toolkit for Interactive Knowledge Discovery

“…The ultimate objective of multi-objective algorithms is to guide the user's decision making, through the provision of a set of solutions that have differing trade-offs between the various objectives [24], and thus the user must be involved in the process of discovering rules . Therefore in the proposed system the user is allowed to control the system by specifying most of the attributes of the system including the rule metrics (objectives), the rule schema, and other parameters as discussed earlier.…”

Evolution of classification rules for comprehensible knowledge discovery

“…Some examples include [1], [2], [3], and [4]. In [2], classification rules are evolved using a Multi-objective Genetic Programming approach.…”

“…Some examples include [1], [2], [3], and [4]. In [2], classification rules are evolved using a Multi-objective Genetic Programming approach. In [4], an approach is presented to discover interesting prediction rules by applying a genetic algorithm in which the adaptive function (fitness function) is divided into two parts.…”