Abstract
Given a variety of classifiers, one prevalent approach in classifier ensemble is to diversely combine classifier components, i.e., diversity-based ensembles, and a lot of previous works show that these ensembles can improve classification accuracy. Random forests are one of the most important ensembles. However, most random forests approaches with diversity-related aspects focus on maximizing tree diversity while producing and training component trees. Alternatively, a novel cognitive-inspired diversity-based random forests method, diversity-based random forests via sample weight learning (DRFS), is proposed. Given numerous component trees from the original random forests, DRFS selects and combines tree classifiers adaptively via diversity learning and sample weight learning. By designing a matrix for the data distribution creatively, a unified optimization model is formulated to learn and select diverse trees, where tree weights are learned through a convex quadratic programming problem with sample weights. Moreover, a self-training algorithm is proposed to solve the convex optimization iteratively and learn sample weights automatically. Comparative experiments on 39 typical UCI classification benchmarks and a variety of real-world text categorization benchmarks of our proposed method are conducted. Extensive experiments show that our method outperforms the traditional methods. Our proposed DRFS method can select and combine tree classifiers adaptively and improves the performance on a variety of classification tasks.
Similar content being viewed by others
Notes
From the view of ensemble pruning, Tsoumakas et al. presented a taxonomy of ensemble pruning methods, i.e., ranking based, clustering based, optimization based, and other categories [45]. Zhou divided related methods into three categories: ordering-based pruning, clustering-based pruning, and optimization-based pruning approaches [53]. More specifically, optimization-based pruning methods formulate the ensemble pruning problem as an optimization problem that aims to find the subset of available component classifiers which maximizes or minimizes an objective related to the generalization ability of the final ensemble, which is also the focus of our paper.
Here, the optimization problem with w and Ω in an iterative learning framework is similar to an EM (expectation-maximization) procedure. We also initially used the EM algorithm in our method, and the results were not encouraged compared with this iterative learning algorithm. However, how to adaptively design, improve, and use a variant of EM algorithm for our DRFS method is a near future topic.
In our experiments, this validation set is bootstrapped from the initial training set, and is used as the new training set for learning sample weights and classifier weights in the iterative learning algorithm for DRFS.
Note that in these experiments with parameters, the used validation set is the same as the validation set in Algorithm 1, and all other experimental conditions are the same as the ones in “Experimental Setup.”
Totally, there are 20 DRFS ensembles with 20 different values λ by 1/λ = {0, 0.1,..., 1, 2,..., 10}.
References
Amasyali MF, Ersoy OK. Classifier ensembles with the extended space forest. IEEE Trans Knowl Data Eng 2014;26(3):549–62.
Amozegar M, Khorasani K. An ensemble of dynamic neural network identifiers for fault detection and isolation of gas turbine engines. Neural Netw 2016;76:106–21.
Ayerdi B, Graṅa M. Hybrid extreme rotation forest. Neural Netw 2014;52:33–42.
Ball K, Grant C, Mundy WR, Shafer TJ. A multivariate extension of mutual information for growing neural networks. Neural Netw 2017;95:29–43.
Bernard S, Adam S, Heutte L. Dynamic random forests. Pattern Recogn Lett 2012;33(12):1580–6.
Biau G. Analysis of a random forests model. J Mach Learn Res 2012;13:1063–95.
Brazdil P, Soares C. A comparison of ranking methods for classification algorithm selection. Proceedings of the 11th European Conference on Machine Learning, pp 63–74; 2000.
Breiman L. Bagging predictors. Mach Learn 1996;24(1):123–40.
Breiman L. Random forests. Mach Learn 2001;45:5–32.
Cardoso-Cachopo A. Improving methods for single-label text categorization. PdD Thesis. Instituto Superior Tecnico: Universidade Tecnica de Lisboa; 2007.
Chang CC, Lin CJ. Libsvm: a library for support vector machines. ACM Trans Intell Syst Technol 2011;2 (3):1–27. http://www.csie.ntu.edu.tw/cjlin/libsvm.
Debole F, Sebastiani F. An analysis of the relative hardness of Reuters-21578 subsets. JASIST 2005;56(6): 584–96. https://doi.org/10.1002/asi.20147.
Demsar J. Statistical comparisons of classifiers over multiple data sets. J Mach Learn Res 2006;7:1–30.
Frank A, Asuncion A. 2010. UCI machine learning repository. University of California, Irvine, School of Information and Computer Sciences. http://archive.ics.uci.edu/ml.
Freund Y, Schapire R. Experiments with a new boosting algorithm. Proceedings of the 13th International Conference on Machine Learning, pp 148–156; 1996.
Freund Y, Schapire R. A decision-theoretic generalization of on-line learning and an application to boosting. J Comput Syst Sci 1997;55(1):119–39.
Hall M, Frank E, Holmes G, Pfahringer B, Reutemann P, Witten IH. The WEKA data mining software: an update. SIGKDD Explor Newsl 2009;11(1):10–8. https://doi.org/10.1145/1656274.1656278.
Han EH, Karypis G. Centroid-based document classification: analysis and experimental results. Principles of Data Mining and Knowledge Discovery, 4th European Conference, PKDD 2000, Lyon, France, September 13-16, 2000, Proceedings, pp 424–431; 2000.
Hansen LK, Salamon P. Neural network ensembles. IEEE Trans Pattern Anal Mach Intell 1990;12(10): 993–1001.
Hothorn T, Hornik K, Zeileis A. Unbiased recursive partitioning: a conditional inference framework. J Comput Graph Stat 2006;15(3):651–74.
Huang K, Zhang R, Jin X, Hussain A. Special issue editorial: cognitively-inspired computing for knowledge discovery. Cogn Comput 2018;10(1):1–2.
Jiang L. Learning random forests for ranking. Frontiers of Computer Science in China 2011;5(1):79–86.
Jiang L, Wang S, Li C, Zhang L. Structure extended multinomial naive Bayes. Inf Sci 2016;329: 346–56.
Krogh A, Sollich P. Statistical mechanics of ensemble learning. Phys Rev E 1997;55(1):811–25.
Kuncheva LI, Whitaker CJ. Measures of diversity in classifier ensembles. Mach Learn 2003;51(2):181–207.
Li N, Yu Y, Zhou ZH. Diversity regularized ensemble pruning. Proceedings of European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases; 2012.
Liu FT, Ting KM. Variable randomness in decision tree ensembles. Advances in Knowledge Discovery and Data Mining, 10th Pacific-Asia Conference, PAKDD 2006, Singapore, April 9-12, 2006, Proceedings, pp 81–90; 2006.
Liu FT, Ting KM, Fan W. Maximizing tree diversity by building complete-random decision trees. Advances in Knowledge Discovery and Data Mining, 9th Pacific-Asia Conference, PAKDD 2005, Hanoi, Vietnam, May 18-20, 2005, Proceedings, pp 605–610; 2005.
Liu FT, Ting KM, Yu Y, Zhou ZH. Spectrum of variable-random trees. J Artif Intell Res (JAIR) 2008;32:355–84.
Lu Z, Wu X, Zhu X, Bongard J. Ensemble pruning via individual contribution ordering. Proceedings of the 16th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Washington, DC, USA, July 25-28, 2010, pp 871–880; 2010.
Lulli A, OnetoEmail L, Anguita D. 2019. Mining big data with random forests. Cognitive Computation pp. 1–23. Published online.
Margineantu D, Dietterich T. Pruning adaptive boosting. Proceedings of International Conference on Machine Learning, pp 211–218; 1997.
Martinez-Munoz G, Hernandez-Lobato D, Suarez A. An analysis of ensemble pruning techniques based on ordered aggregation. IEEE Trans Pattern Anal Mach Intell 2009;31(2):245–59.
McCallum A, Nigam K. 1998. A comparison of event models for naive Bayes text classification. In: Learning for text categorization: papers from the 1998 AAAI Workshop, pp 41–48. http://www.kamalnigam.com/papers/multinomial-aaaiws98.pdf.
Menze BH, Kelm BM, Splitthoff DN, Kothe U, Hamprecht FA. On oblique random forests. Proceedings of European Conference on Machine Learning and Knowledge Discovery in Databases, (ECML-PKDD’11), pp 453–469; 2011.
Opitz DW, Shavlik JW. Generating accurate and diverse members of a neural network ensemble. Advances in Neural Information Processing Systems (NIPS’96), pp 535–541. MIT Press; 1996.
Osadchy M, Keren D, Raviv D. Recognition using hybrid classifiers. IEEE Trans Pattern Anal Mach Intell 2016;38(4):759–71.
Perera AG, Law YW, Chahl JS. Human pose and path estimation from aerial video using dynamic classifier selection. Cogn Comput 2018;10(6):1019–41.
Qiu C, Jiang L, Li C. Randomly selected decision tree for test-cost sensitive learning. Appl Soft Comput 2017;53:27–33.
Quinlan JR. 1993. C4.5: Programs for machine learning. Morgan Kaufmann.
Robnik-Sikonja M. Improving random forests. Proceedings of 15th European Conference on Machine Learning (ECML’04), pp 359–370; 2004.
Rodriguez JJ, Kuncheva LI, Alonso CJ. Rotation forest: a new classifier ensemble method. IEEE Trans Pattern Anal Mach Intell 2006;28(10):1619–30.
Tang B, He H, Baggenstoss PM, Kay S. A Bayesian classification approach using class-specific features for text categorization. IEEE Trans Knowl Data Eng 2016;28(6):1602–06.
Trawinski K, Quirin A, Cordon O. On the combination of accuracy and diversity measures for genetic selection of bagging fuzzy rule-based multiclassification systems. Proceedings of the 9th Intelligent Systems Design and Applications, pp 121–127; 2009.
Tsoumakas G, Partalas I, Vlahavas I. An ensemble pruning primer. Applications of Supervised and Unsupervised Ensemble Methods, pp 1–13; 2009.
Wen G, Hou Z, Li H, Li D, Jiang L, Xun E. Ensemble of deep neural networks with probability-based fusion for facial expression recognition. Cogn Comput 2017;9(5):597–610.
Wolpert D. Stacked generalization. Neural Netw 1992;5(2):241–60.
Yang C, Yin XC, Hao HW. Diversity-based ensemble with sample weight learning. 22nd International Conference on Pattern Recognition, ICPR 2014, Stockholm, Sweden, August 24-28, 2014, pp 1236–1241; 2014.
Yang C, Yin XC, Huang K. Text categorization with diversity random forests. Neural Information Processing - 21st International Conference, ICONIP 2014, Kuching, Malaysia, November 3-6, 2014. Proceedings, Part III, pp 317–324; 2014.
Yin XC, Huang K, Hao HW, Iqbal K, Wang ZB. A novel classifier ensemble method with sparsity and diversity. Neurocomputing 2014;134:214–21.
Yin XC, Huang K, Yang C, Hao HW. Convex ensemble learning with sparsity and diversity. Inf Fusion 2014;20:49–59.
Zhang Y, Burer A, Street WN, Bennett K, Parrado-hern E. Ensemble pruning via semi-definite programming. J Mach Learn Res 2006;7:1315–38.
Zhou ZH. Ensemble methods: foundations and algorithms. Boca Raton: Chamman & Hall/CRC; 2012.
Zhou ZH, Wu J, Tang W. Ensembling neural networks: many could be better than all. Artif Intell 2002; 137:239–63.
Funding
The research was partly supported by the National Natural Science Foundation of China (61473036), China Postdoctoral Science Foundation (2018M641199), and Beijing Natural Science Foundation (4194084).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Ethical Approval
This article does not contain any studies with human participants or animals performed by any of the authors.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Yang, C., Yin, XC. Diversity-Based Random Forests with Sample Weight Learning. Cogn Comput 11, 685–696 (2019). https://doi.org/10.1007/s12559-019-09652-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12559-019-09652-0