Abstract
This paper introduces a method to rebalance the output of classification algorithms using the corresponding confusion matrices. This is done by modifying the classification output, i.e. reweighting predictions, when they can be interpreted as probabilities. The method is evaluated and analyzed via experiments involving a number of classifiers and both standard and real life datasets. Our results show that confusion matrix based reweighting can be used to achieve certain kinds of balance in classification, while maintaining the same level of accuracy.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
References
Garcia, et al.: The class imbalance problem in pattern classification and learning. Congreso Espanol de Informatica, 283–291 (2007)
Galar, et al.: A Review on Ensembles for the Class Imbalance Problem: Bagging-, Boosting-, and Hybrid-Based Approaches. IEEE Transactions on Systems, Man, and Cybernetics, Part C, 463–484 (2012)
Guo, et al.: On the Class Imbalance Problem. In: Proceedings of the 2008 Fourth International Conference on Natural Computation, vol. 04, pp. 192–201 (2008)
Elkan: The foundations of cost-sensitive learning. In: International Joint Conference on Artificial Intelligence, pp. 973–978 (2001)
Garcia, et al.: Exploring the Performance of Resampling Stategies for the Class Imbalance Problem. In: 23rd International Conference on Industrial Engineering and Other Applications of Applied Intelligent Systems, pp. 541–549 (2010)
Orriols-Puig, et al.: The class imbalance problem in learning classifier systems: a preliminary study. In: GECOO Workshops, pp. 74–78 (2005)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2013 Springer International Publishing Switzerland
About this paper
Cite this paper
Warmerdam, V.D., Szlávik, Z. (2013). Confusion Matrix Based Reweighting. In: Ali, M., Bosse, T., Hindriks, K., Hoogendoorn, M., Jonker, C., Treur, J. (eds) Contemporary Challenges and Solutions in Applied Artificial Intelligence. Studies in Computational Intelligence, vol 489. Springer, Heidelberg. https://doi.org/10.1007/978-3-319-00651-2_19
Download citation
DOI: https://doi.org/10.1007/978-3-319-00651-2_19
Publisher Name: Springer, Heidelberg
Print ISBN: 978-3-319-00650-5
Online ISBN: 978-3-319-00651-2
eBook Packages: EngineeringEngineering (R0)