ABSTRACT
A priori determining the ideal number of component classifiers of an ensemble is an important problem. The volume and velocity of big data streams make this even more crucial in terms of prediction accuracies and resource requirements. There is a limited number of studies addressing this problem for batch mode and none for online environments. Our theoretical framework shows that using the same number of independent component classifiers as class labels gives the highest accuracy. We prove the existence of an ideal number of classifiers for an ensemble, using the weighted majority voting aggregation rule. In our experiments, we use two state-of-the-art online ensemble classifiers with six synthetic and six real-world data streams. The violation of providing independent component classifiers for our theoretical framework makes determining the exact ideal number of classifiers nearly impossible. We suggest upper bounds for the number of classifiers that gives the highest accuracy. An important implication of our study is that comparing online ensemble classifiers should be done based on these ideal values, since comparing based on a fixed number of classifiers can be misleading.
- X. Zhu, Stream Data Mining Repository, http://www.cse.fau.edu/ xqzhu/stream.html, 2010.Google Scholar
- E. Bauer and R. Kohavi. An empirical comparison of voting classification algorithms: Bagging, boosting, and variants. Mach. Learn., 36(1-2):105--139, 1999. Google ScholarDigital Library
- A. Bifet, G. Holmes, and B. Pfahringer. Leveraging bagging for evolving data streams. In ECML PKDD, pages 135--150, 2010. Google ScholarDigital Library
- A. Bifet, G. Holmes, B. Pfahringer, R. Kirkby, and R. Gavalda. New ensemble methods for evolving data streams. In ACM SIGKDD, pages 139--148, 2009. Google ScholarDigital Library
- D. Brzezinski and J. Stefanowski. Reacting to different types of concept drift: The accuracy updated ensemble algorithm. IEEE TNNLS, 25(1):81--94, 2014.Google Scholar
- L.-W. Chan. Weighted least square ensemble networks. In IJCNN, volume 2, pages 1393--1396, 1999.Google Scholar
- G. Fumera and F. Roli. A theoretical and experimental analysis of linear combiners for multiple classifier systems. IEEE TPAMI, 27(6):942--956, 2005. Google ScholarDigital Library
- G. Fumera, F. Roli, and A. Serrau. A theoretical analysis of bagging as a linear combination of classifiers. IEEE TPAMI, 30(7):1293--1299, 2008. Google ScholarDigital Library
- P. C. Hansen, V. Pereyra, and G. Scherer. Least Squares Data Fitting with Applications. JHU Press, 2013.Google Scholar
- D. Hernandez-Lobato, G. Martinez-Munoz, and A. Suarez. How large should ensembles of classi ers be? Patt. Recog., 46(5):1323--1336, 2013. Google ScholarDigital Library
- L. I. Kuncheva and C. J. Whitaker. Measures of diversity in classifier ensembles and their relationship with the ensemble accuracy. Mach. Learn., 51(2):181--207, 2003. Google ScholarDigital Library
- P. Latinne, O. Debeir, and C. Decaestecker. Limiting the number of trees in random forests. In MCS, pages 178--187, 2001. Google ScholarDigital Library
- T. M. Oshiro, P. S. Perez, and J. A. Baranauskas. How many trees in a random forest? In MLDM, pages 154--168, 2012. Google ScholarDigital Library
- N. C. Oza and S. J. Russell. Experimental comparisons of online and batch versions of bagging and boosting. In ACM SIGKDD, pages 359--364, 2001. Google ScholarDigital Library
Index Terms
- A Theoretical Framework on the Ideal Number of Classifiers for Online Ensembles in Data Streams
Recommendations
How large should ensembles of classifiers be?
We propose to determine the size of a parallel ensemble by estimating the minimum number of classifiers that are required to obtain stable aggregate predictions. Assuming that majority voting is used, a statistical description of the convergence of the ...
Optimal ensemble construction via meta-evolutionary ensembles
In this paper, we propose a meta-evolutionary approach to improve on the performance of individual classifiers. In the proposed system, individual classifiers evolve, competing to correctly classify test points, and are given extra rewards for getting ...
A geometric framework for multiclass ensemble classifiers
AbstractEnsemble classifiers have been investigated by many in the artificial intelligence and machine learning community. Majority voting and weighted majority voting are two commonly used combination schemes in ensemble learning. However, understanding ...
Comments