ABSTRACT
In existing artificial classification systems, the problem domain is created and controlled by humans. Humans set up and tune the problem domain, such as determining the problem's complexity. If humans can set up the problem appropriately then the machines can extract beneficial knowledge to solve classification task. This paper introduces an autonomous classification problem generation approach. The classification problem's difficulty is adapted based on the classification agent's performance within the defined attributes. An automated problem generator has been created to evolve the simulated datasets whilst the classification agent, in this case a learning classifier system, attempts to learn the evolving problem. The idea here is to tune the datasets autonomously such that the problem characteristics may be determined efficiently to empirically test the learning bounds of the classification agent by lowering human involvement. In this way, the effect of the problem's characteristics, which alter the classification agent's performance, becomes human readable. Tabu Search has been applied in the problem generator to discover the best combination of domain features in order to adjust the problem's complexity. Experiments confirm that the problem generator was able to tune the problem's complexity either to make the problem 'harder' or 'easier' so that it can either 'increase' or 'decrease' the classification agent's performance.
- C. M. Bishop. Pattern Recognition and Machine Learning. Springer, Natural Computing Series. 2006. Google ScholarDigital Library
- R. O. Duda, P. E. Hart and D. G. Stork. Pattern Classification 2nd Edition. John Wiley, 2001. Google ScholarDigital Library
- Tin Kam Ho and Mitra Basu. Complexity Measure of Supervised Classification Problems. IEEE Transactions on Pattern Analysis and Machine Intelligence, Volume 24/2002(3):289--300. 2002. Google ScholarDigital Library
- M. V. Butz. Kernel-based, Ellipsoidal Conditions in the Real-Valued XCS Classifier System. Genetic Evolutionary Computational Conference (GECCO 2005). ACM, 2005. Google ScholarDigital Library
- M.V. Butz. Rule-Based Evolutionary Online Learning Systems: A Principal Approach to LCS Analysis and Design. Springer. 2006.Google Scholar
- Muhammad Iqbal, Will N. Browne and Mengjie Zhang. XCSR with Computed Continuous Action. Australisian AI 2012, pages 350--361. Springer, 2012. Google ScholarDigital Library
- F. Kharbat, L. Bull and M. Odeh. Revisiting Genetic Selection in the XCS Learning Classifier System. The 2005 IEEE Congress on Evolutionary Computation, pages 2061--2068. 2005.Google ScholarCross Ref
- T. Kovacs. Strength or Accuracy: Credit Assignment in Learning Classifier Systems. Springer, 2004. Google ScholarDigital Library
- Syahaneim Marzukhi, Will N. Browne and Mengjie Zhang. Two-Cornered Learning Classifier Systems for Pattern Generation and Classification. The 12th Genetic and Evolutionary Computation Conference (GECCO 2012), pages 895--902. ACM, 2012. Google ScholarDigital Library
- R. J. Urbanowicz and J. H. Moore. Review Article Learning Classifier Systems: A Complete Introduction, Review, and Roadmap. Journal of Artificial Evolution and Applications, Volume 2009:1--25. 2009. Google ScholarCross Ref
- O. Sigaud and S. W. Wilson. Learning Classifier Systems: A Survey. Soft Computing - A Fusion of Foundations, Methodologies and Applications, Volume 11(11):1065--1078. 2007. Google ScholarDigital Library
- S. W. Wilson. Classifier Fitness Based on Accuracy. Evolutionary Computation, Volume 3 (2):149--175. Massachusetts Institute of Technology, 1995. Google ScholarDigital Library
- S. W. Wilson. Get Real! XCS with Continuous-Valued Inputs. Learning Classifier Systems, From Foundations to Applications, LNAI-1813,pages 209--219. Springer, 2000. Google ScholarDigital Library
- S. W. Wilson. Coevolution of Pattern Generators and Recognizers. Lecture Notes in Computer Science (LNCS), Volume 6471/2010(1):38--46. 2010.Google ScholarCross Ref
- Nuria Macia, Albert Orriols-Puig and Ester Bernado-Mansilla. Beyond Homemade Artificial Data Sets. Hybrid Artificial Intelligence Systems Lecture Notes in Computer Science, Volume 5572/2009:605--612. 2009. Google ScholarDigital Library
- Z. Michalewicz and D. B. Fogel. How to Solve It: Modern Heuristics. Springer, 2000. Google ScholarDigital Library
- J. H. Holland. Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, Artificial Intelligence. University of Michigan Press, pages 313--329. 1975.Google Scholar
- Sean Luke. Essentials of Metaheuristics: A Set of Undergraduate Lecture Notes. Department of Computer Science, George Mason University. 2010.Google Scholar
- Micheal Gendreau. A Tutorial On the Tabu Search. Department of Computer Science, de Montreal University, Canada. 2000.Google Scholar
Index Terms
- Adaptive artificial datasets through learning classifier systems for classification tasks
Recommendations
Three-cornered coevolution learning classifier systems for classification tasks
GECCO '14: Proceedings of the 2014 Annual Conference on Genetic and Evolutionary ComputationThe Three-Cornered Coevolution concept describes a framework where artificial problems may be generated in concert with classification agents in order to provide insight into their relationships. This is unlike standard studies where humans set a ...
Adaptive artificial datasets to discover the effects of domain features for classification tasks
GECCO '13 Companion: Proceedings of the 15th annual conference companion on Genetic and evolutionary computationThis paper described an automated pattern generator to generate various synthetic data sets for classification problems, where the problem's complexity can be manipulated autonomously. The Tabu Search technique has been applied in the pattern generator ...
Accuracy-based learning classifier systems: models, analysis and applications to classification tasks
Recently, Learning Classifier Systems (LCS) and particularly XCS have arisen as promising methods for classification tasks and data mining. This paper investigates two models of accuracy-based learning classifier systems on different types of ...
Comments