Abstract
We study an approach for speeding up the training of error-correcting output codes (ECOC) classifiers. The key idea is to avoid unnecessary computations by exploiting the overlap of the different training sets in the ECOC ensemble. Instead of re-training each classifier from scratch, classifiers that have been trained for one task can be adapted to related tasks in the ensemble. The crucial issue is the identification of a schedule for training the classifiers which maximizes the exploitation of the overlap. For solving this problem, we construct a classifier graph in which the nodes correspond to the classifiers, and the edges represent the training complexity for moving from one classifier to the next in terms of the number of added training examples. The solution of the Steiner Tree problem is an arborescence in this graph which describes the learning scheme with the minimal total training complexity. We experimentally evaluate the algorithm with Hoeffding trees, as an example for incremental learners where the classifier adaptation is trivial, and with SVMs, where we employ an adaptation strategy based on adapted caching and weight reuse, which guarantees that the learned model is the same as per batch learning.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Allwein, E.L., Schapire, R.E., Singer, Y.: Reducing multiclass to binary: A unifying approach for margin classifiers. J. Mach. Learn. Res (JMLR) 1, 113–141 (2000)
Bifet, A., Holmes, G., Kirkby, R., Pfahringer, B.: MOA: Massive Online Analysis. J. Mach. Learn. Res., JMLR (2010), http://sourceforge.net/projects/moa-datastream/
Blockeel, H., Struyf, J.: Efficient algorithms for decision tree cross-validation. J. Mach. Learn. Res. (JMLR) 3, 621–650 (2003)
Chang, C.-C., Lin, C.-J.: LIBSVM: a library for support vector machines (2001), Software, available at http://www.csie.ntu.edu.tw/~cjlin/libsvm
Dietterich, T.G., Bakiri, G.: Solving multiclass learning problems via error-correcting output codes. J. Artif. Intell. Res. (JAIR) 2, 263–286 (1995)
Domingos, P., Hulten, G.: Mining high-speed data streams. In: KDD, Boston, MA, USA, pp. 71–80. ACM, New York (2000)
Frank, A., Asuncion, A.: UCI machine learning repository (2010)
Friedman, J.H.: Another approach to polychotomous classification. Technical report, Department of Statistics, Stanford University, Stanford, CA (1996)
Fürnkranz, J.: Round robin classification. J. Mach. Learn. Res. (JMLR) 2, 721–747 (2002)
Joachims, T.: Making large-scale SVM learning practical. In: Schölkopf, B., Burges, C., Smola, A. (eds.) Advances in Kernel Methods - Support Vector Learning, pp. 169–184. MIT Press, Cambridge (1999)
Park, S.-H., Fürnkranz, J.: Efficient decoding of ternary error-correcting output codes for multiclass classification. In: Buntine, W.L., Grobelnik, M., Mladenić, D., Shawe-Taylor, J. (eds.) ECML/PKDD-09, Part II, Bled, Slovenia, pp. 189–204. Springer, Heidelberg (2009)
Park, S.-H., Weizsäcker, L., Fürnkranz, J.: Exploiting code-redundancies in ECOC for reducing its training complexity using incremental and SVM learners. Technical Report TUD-KE-2010-06, TU Darmstadt (July 2010)
Pimenta, E., Gama, J., Carvalho, A.: Pursuing the best ecoc dimension for multiclass problems. In: Wilson, D., Sutcliffe, G. (eds.) FLAIRS Conference, pp. 622–627. AAAI Press, Menlo Park (2007)
Wong, R.: A dual ascent approach for steiner tree problems on a directed graph. Mathematical Programming 28(3), 271–287 (1984)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2010 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Park, SH., Weizsäcker, L., Fürnkranz, J. (2010). Exploiting Code Redundancies in ECOC. In: Pfahringer, B., Holmes, G., Hoffmann, A. (eds) Discovery Science. DS 2010. Lecture Notes in Computer Science(), vol 6332. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-16184-1_19
Download citation
DOI: https://doi.org/10.1007/978-3-642-16184-1_19
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-16183-4
Online ISBN: 978-3-642-16184-1
eBook Packages: Computer ScienceComputer Science (R0)