A bi-phased multi-objective genetic algorithm based classifier. (15th May 2020)

• Record Type: Journal Article
• Authors: Dutta, Dipankar; Sil, Jaya; Dutta, Paramartha
• Is Part Of: Expert systems with applications. Volume 146(2020)
• Publication Date: 2020-05-15
• Subjects: Classification rules mining -- Elitist Multi-Objective Genetic Algorithm -- Pareto approach -- Statistical test; ADI Adaptive Discretization Intervals -- ARri_max Maximum value of rth feature gene in ith chromosome -- ARri_min Minimum value of rth feature gene in ith chromosome -- BioHEL Bioinformatics-oriented Hierarchical Evolutionary Learning -- BPMOGA Bi-Phased Multi-Objective Genetic Algorithm -- Cnf Confidence -- CNN Center Nearest Neighbor -- CORE CO-evolutionary Rule Extractor -- Cov Coverage -- CR Complete Rule -- CSR Class Specific -- EFS−RPS Evolutionary Feature Selection for fuzzy Rough set based Prototype Selection -- GA Genetic Algorithm -- HIDER HIerarchical DEcision Rules -- ILGA Incremental Learning with GAs -- IP Initial Population -- KEEL Knowledge Extraction Evolutionary Learning -- LCS Learning Classifier System -- max(Ai) Largest value of ith feature of the training data set -- min(Ai) Smallest value of ith feature of the training data set -- MLP−BP Multilayer Perceptron with Backpropagation -- MOGA Multi-Objective GA -- MOOP Multi-Objective Optimization Problem -- NB Nai¨ve Bayes -- NCSR Number of CSRs -- NOC Number of Cover -- NOM Number of Match -- NSGA−II Non-dominated Sorting GA-II -- NSLV New Structural Learning algorithm in Vague environment -- NVF Number of Valid Features -- SUP(Ant∧Cons) Support of the antecedent and consequence -- SUP(Ant) Support of the antecedent -- SUP(Cons) Support of the consequence -- SWOT Strengths, Weaknesses, Opportunities, and Threats -- TCnf Total Confidence -- TCov Total Coverage -- UCI University of California at Irvine -- UCS Upervised Classifier System -- 10-CV 10 fold Cross Validation; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2019.113163
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: We have proposed a Bi-Phased Multi-Objective Genetic Algorithm. We have done data classification by the proposed algorithm. It is a cyclic algorithm. Statistical tests show that the performance of the proposed algorithm is comparable. Abstract: This paper presents a novel Bi-Phased Multi-Objective Genetic Algorithm ( BPMOGA ) based classification method. It is a Learning Classifier System ( LCS ) designed for supervised learning tasks. Here we have used Genetic Algorithms ( GA s) to discover optimal classifiers from data sets. The objective of the work is to find out a classifier or Complete Rule ( CR ) which comprises of several Class Specific Rules ( CSR s). Phase-I of BPMOGA extracts optimized CSR s in I F − T H E N form by following Michigan approach, without considering interaction among the rules. Phase-II of BPMOGA builds optimized CR s from CSR s by following Pittsburgh way. It combines the advantages of both approaches. Extracted CR s help to build CSR s for the next run of phase-I. Hence, phase-I and phase-II are cyclically related, which is one of the uniqueness of BPMOGA . With the help of twenty one benchmark data sets from the University of California at Irvine ( UCI ) machine learning repository we have compared performance of BPMOGA based classifier with fourteen GA and non- GA based classifiers. Statistical test shows that the performance of the proposed classifier is either superior or comparable to other classifiers.