Rule-based machine learning classification and knowledge discovery for complex problems

Abstract

Learning classifier systems (LCSs) are an advantageous, powerful, and flexible class of algorithms that have, to date, been underutilized largely due to the perception that they are difficult to apply, evaluate, and interpret. ExSTraCS is an Extended Supervised Tracking and Classifying System based on the Michigan-Style LCS architecture [4]. It offers an accessible, user friendly LCS platform for supervised rule-based machine learning, classification, data mining, prediction, and knowledge discovery. ExSTraCS seeks to make no assumptions about the data, and is therefore model free and particularly well suited to complex problems that are multi-factorial, interacting (non-linear), heterogeneous, noisy, class imbalanced, multi-class, or larger-scale. ExSTraCS is written in Python, open source, well documented, and freely available at sourceforge.net.

References

1. Jaume Bacardit and Natalio Krasnogor. A mixed discrete-continuous attribute list representation for large scale classification domains. In Proceedings of the 11th Annual conference on Genetic and evolutionary computation, pages 1155--1162. ACM, 2009.
2. Ester Bernadó-Mansilla and Josep M. Garrell-Guiu. Accuracy-based learning classifier system: models, analysis and applications to classification tasks. Evolutionary Computation, 11(3):209--238, 2003.
3. Jie Tan, Jason Moore, and Ryan Urbanowicz. Rapid rule compaction strategies for global knowledge discovery in a supervised learning classifier system. In Advances in Artificial Life, ECAL, volume 12, pages 110--117, 2013.
4. Ryan Urbanowicz, Gediminas Bertasius, and Jason Moore. An extended Michigan-style learning classifier system for flexible supervised learning, classification, and data mining. Parallel Problem Solving from Nature--PPSN XIII, page In press, 2014.
5. Ryan Urbanowicz and Jason Moore. Addressing scalability with a rule specificity limit in a Michigan-style supervised learnign classifier system for classification, prediction, and knowledge discovery. In Review, 2014.
6. Ryan Urbanowicz, Ambrose Granizo-Mackenzie, and Jason Moore. Instance-linked attribute tracking and feedback for Michigan-style supervised learning classifier systems. In Proceedings of the fourteenth international conference on Genetic and evolutionary computation conference, pages 927--934. ACM, 2012.
7. Ryan J Urbanowicz, Ambrose Granizo-Mackenzie, and Jason H Moore. An analysis pipeline with statistical and visualization-guided knowledge discovery for Michigan-style learning classifier systems. Computational Intelligence Magazine, IEEE, 7(4):35--45, 2012.
8. Ryan J Urbanowicz, Delaney Granizo-Mackenzie, and Jason H Moore. Using expert knowledge to guide covering and mutation in a Michigan-style learning classifier system to detect epistasis and heterogeneity. In Parallel Problem Solving from Nature-PPSN XII, pages 266--275. Springer, 2012.
9. Stewart W. Wilson. Classifier fitness based on accuracy. Evolutionary computation, 3(2):149--175, 1995.

Index Terms

• Computing methodologies

  • Artificial intelligence

    • Knowledge representation and reasoning

  • Machine learning

    • Learning paradigms

      • Supervised learning

        • Supervised learning by classification

    • Machine learning approaches

      • Classification and regression trees

• Theory of computation

  • Logic