ABSTRACT
Traditional accuracy-based XCS classifier system generally learns and evolves classifiers from scratch when facing each particular problem. Inspired by humans with the ability to learn new skills by inducing knowledge from related problems, transfer learning (TL) focuses on leveraging the knowledge of source domains to help the problem solving of another different but related domain. This paper attempts to combine XCS and TL to propose a novel extension transferable XCS (tXCS). tXCS utilizes the inherent characteristics of XCS, that naturally discovers expressive classifiers as the generalized knowledge of domains, to realize the classifier transfer from source domains to a target domain that makes it learn faster, which is conceptually different from the previous integrations between XCS and TL. The systematic study is presented to verify the ability of knowledge transfer between domains with different degrees of similarity, which has been pointed out to be the challenge of TL. We demonstrate that tXCS can significantly speed up the learning efficiency of canonical XCS in both of single-step and multi-step benchmark problems.
- I. M. Alvarez, W. N. Browne, and M. Zhang. Reusing learned functionality to address complex boolean functions. In SEAL, pages 383--394, 2014. Google ScholarDigital Library
- J. Bacardit and M. Butz. Data mining in learning classifier systems: Comparing XCS with GAssist. In Learning Classifier Systems, pages 282--290. 2007. Google ScholarDigital Library
- K. Bache and M. Lichman. UCI machine learning repository, 2013.Google Scholar
- E. Bernado-Mansilla and J. M. Garrell-Guiu. Accuracy-based learning classifier systems: models, analysis and applications to classification tasks. Evol. Comput., 11(3):209--238, 2003. Google ScholarDigital Library
- M. V. Butz. XCSJava 1.0: An implementation of the XCS classifier system in Java. Technical Report No. 2000027, Illinois Genetic Algorithms Laboratory, UIUC, 2000.Google Scholar
- M. V. Butz, D. E. Goldberg, and P. L. Lanzi. Gradient descent methods in learning classifier systems: Improving XCS performance in multistep problems. IEEE Trans. on Evol. Comput., 9(5):452--473, 2005. Google ScholarDigital Library
- M. V. Butz, T. Kovacs, P. L. Lanzi, and S. W. Wilson. Toward a theory of generalization and learning in XCS. IEEE Trans. on Evol. Comput., 8(1):28--46, 2004. Google ScholarDigital Library
- M. V. Butz and M. Pelikan. Studying XCS/BOA learning in boolean functions: structure encoding and random boolean functions. In Proceedings of the annual conference on Genetic and Evol. Comput., pages 1449--1456, 2006. Google ScholarDigital Library
- M. V. Butz and S. W. Wilson. An algorithmic description of XCS. Soft Computing, 6(3-4):144--153, 2002.Google ScholarCross Ref
- W. Dai, Q. Yang, G.-R. Xue, and Y. Yu. Boosting for transfer learning. In Proceedings of the international conference on Machine learning, pages 193--200, 2007. Google ScholarDigital Library
- S. Dzeroski, L. De Raedt, and K. Driessens. Relational reinforcement learning. Machine Learning, 43(1-2):7--52, 2001.Google ScholarDigital Library
- L. Feng, Y. Ong, M. Lim, and I. W. Tsang. Memetic search with interdomain learning: A realization between CVRP and CARP. IEEE Trans. Evolutionary Computation, 19(5):644--658, 2015.Google ScholarCross Ref
- M. Iqbal, W. Browne, and M. Zhang. Reusing building blocks of extracted knowledge to solve complex, large-scale boolean problems. IEEE Trans. on Evol. Comput., 18(4):465--480, 2014.Google ScholarCross Ref
- G. Konidaris, I. Scheidwasser, and A. Barto. Transfer in reinforcement learning via shared features. Journal of Machine Learning Research, 13:1333--1371, 2012. Google ScholarDigital Library
- T. Kovacs. XCS classifier system reliably evolves accurate, complete, and minimal representations for boolean functions. In Soft comput. in eng. design and manufacturing, pages 59--68. Springer, 1998.Google Scholar
- T. Kovacs. Genetics-based machine learning. In Handbook of Natural Computing, pages 937--986. Springer, 2012.Google ScholarCross Ref
- P. L. Lanzi. An analysis of generalization in the XCS classifier system. Evol. Comput., 7(2):125--149, 1999. Google ScholarDigital Library
- P. L. Lanzi, W. Stolzmann, and S. W. Wilson. Learning classifier systems: from foundations to applications. Springer, 2000. Google ScholarDigital Library
- P. Larra~naga and J. A. Lozano. Estimation of Distribution Algorithms. A New Tool for Evolutionary Computation. Kluwer Academic Publishers, 2002. Google ScholarDigital Library
- X. Li, S. Mabu, and K. Hirasawa. A novel graph-based estimation of the distribution algorithm and its extension using reinforcement learning. IEEE Trans. Evol. Comput., 18(1):98--113, 2014.Google ScholarCross Ref
- D. Mellor. Policy transfer with a relational learning classifier system. In Proceedings of the workshops on Genetic and Evol. Comput., pages 82--84, 2005. Google ScholarDigital Library
- S. J. Pan and Q. Yang. A survey on transfer learning. IEEE Trans. on Knowledge and Data Engineering, 22(10):1345--1359, 2010. Google ScholarDigital Library
- M. Pelikan, M. Hauschild, and P. Lanzi. Transfer learning, soft distance-based bias, and the hierarchical BOA. In Parallel Problem Solving from Nature - PPSN XII, pages 173--183. 2012. Google ScholarDigital Library
- R. Santana, A. Mendiburu, and J. A. Lozano. Structural transfer using EDAs: An application to multi-marker tagging snp selection. In IEEE Congr. on Evol. Comput., pages 3221--3228, 2012.Google ScholarCross Ref
- X. Shi, Q. Liu, W. Fan, and P. Yu. Transfer across completely different feature spaces via spectral embedding. IEEE Trans. on Knowledge and Data Engineering, 25(4):906--918, 2013. Google ScholarDigital Library
- O. Sigaud and S. W. Wilson. Learning classi er systems: a survey. Soft Computing, 11(11):1065--1078, 2007. Google ScholarDigital Library
- K. O. Stanley, D. B. D'Ambrosio, and J. Gauci. A hypercube-based encoding for evolving large-scale neural networks. Artif. Life, 15(2):185--212, 2009. Google ScholarDigital Library
- R. S. Sutton and A. G. Barto. Reinforcement Learning: An Introduction. MIT Press, 1998. Google ScholarDigital Library
- M. E. Taylor and P. Stone. Cross-domain transfer for reinforcement learning. In Proceedings of the international conference on Machine learning, pages 879--886, 2007. Google ScholarDigital Library
- M. E. Taylor and P. Stone. Transfer learning for reinforcement learning domains: A survey. Journal of Machine Learning Research, 10:1633--1685, 2009. Google ScholarDigital Library
- R. J. Urbanowicz and J. H. Moore. Learning classifier systems: a complete introduction, review, and roadmap. Journal of Artificial Evolution and Applications, 2009:1--25, 2009. Google ScholarCross Ref
- P. Verbancsics and K. O. Stanley. Evolving static representations for task transfer. Journal of Machine Learning Research, 11:1737--1769, 2010. Google ScholarDigital Library
- C. J. C. H. Watkins. Learning from delayed rewards. PhD thesis, University of Cambridge, 1989.Google Scholar
- S. W. Wilson. ZCS: A zeroth level classifier system. Evol. Comput., 2(1):1--18, 1994. Google ScholarDigital Library
- S. W. Wilson. Classifier tness based on accuracy. Evol. Comput., 3(2):149--175, 1995. Google ScholarDigital Library
- S. W. Wilson. Mining oblique data with XCS. In Advances in Learning Classifier Systems, pages 158--174. 2001. Google ScholarDigital Library
Index Terms
- Transferable XCS
Recommendations
Towards Adapting XCS for Imbalance Problems
PRICAI '08: Proceedings of the 10th Pacific Rim International Conference on Artificial Intelligence: Trends in Artificial IntelligenceThe class imbalance problem has been recognized as a crucial problem in machine learning and data mining. Learning systems tend to be biased towards the majority class and thus have poor performance in classifying the minority class instances. This ...
To handle real valued input in XCS: using fuzzy hyper-trapezoidal membership in classifier condition
SEAL'10: Proceedings of the 8th international conference on Simulated evolution and learningLearning classifier systems (LCSs) are evolutionary learning mechanisms that combine genetic algorithms (GAs) with the power of the reinforcement learning paradigm. XCS, eXtended Classifier System, is currently considered as state of the art learning ...
XCS-CR: determining accuracy of classifier by its collective reward in action set toward environment with action noise
GECCO '18: Proceedings of the Genetic and Evolutionary Computation Conference CompanionAccuracy based Learning Classifier System (XCS) prefers to generalize the classifiers that always acquire the same reward, because they make accurate reward predictions. However, real-world problems have noise, which means that classifiers may not ...
Comments