Abstract
Significant changes in the instance distribution or associated cost function of a learning problem require one to reoptimize a previously-learned classifier to work under new conditions. We study the problem of reoptimizing a multi-class classifier based on its ROC hypersurface and a matrix describing the costs of each type of prediction error. For a binary classifier, it is straightforward to find an optimal operating point based on its ROC curve and the relative cost of true positive to false positive error. However, the corresponding multi-class problem (finding an optimal operating point based on a ROC hypersurface and cost matrix) is more challenging and until now, it was unknown whether an efficient algorithm existed that found an optimal solution. We answer this question by first proving that the decision version of this problem is \(\mathsf{NP}\) -complete. As a complementary positive result, we give an algorithm that finds an optimal solution in polynomial time if the number of classes n is a constant. We also present several heuristics for this problem, including linear, nonlinear, and quadratic programming formulations, genetic algorithms, and a customized algorithm. Empirical results suggest that under both uniform and non-uniform cost models, simple greedy methods outperform more sophisticated methods.
Article PDF
Similar content being viewed by others
References
Abramson, M. A. (2005). Genetic algorithm and direct search toolbox. http://www.mathworks.com/.
Blake, C., & Merz, C. (2005). UCI repository of machine learning databases. http://www.ics.uci.edu/~mlearn/MLRepository.html.
Boyd, S., & Vandenberghe, L. (2004). Convex optimization. Cambridge: Cambridge University Press.
Demšar, J. (2006). Statistical comparisons of classifiers over multiple data sets. Journal of Machine Learning Research, 7, 1–30.
Deng, K., Bourke, C., Scott, S. D., & Vinodchandran, N. V. (2006). New algorithms for optimizing multi-class classifiers via ROC surfaces. In Proceedings of the ICML workshop on ROC analysis in machine learning (pp. 17–24).
Ferri, C., Hernández-Orallo, J., & Salido, M. (2003). Volume under the ROC surface for multi-class problems. In European conference on artificial intelligence (pp. 108–120).
Fieldsend, J., & Everson, R. (2005). Formulation and comparison of multi-class ROC surfaces. In Proceedings of the ICML workshop on ROC analysis in machine learning (pp. 41–48).
Grant, M., Boyd, S., & Ye, Y. (2006). Disciplined convex programming. In L. Liberti & N. Maculan (Eds.), Nonconvex Optimization and its Applications. Global optimization: From theory to implementation (pp. 155–210). Berlin: Springer. Available at http://www.stanford.edu/~boyd/cvx/.
Hand, D., & Till, R. (2001). A simple generalisation of the area under the ROC curve for multiple class classification problems. Machine Learning, 45, 171–186.
Hochberg, Y., & Tamhane, A. C. (1987). Multiple comparison procedures. New York: Wiley.
Kohli, R., Krishnamurti, R., & Mirchandani, P. (1994). The minimum satisfiability problem. SIAM Journal of Discrete Mathematics, 7, 275–283.
Lachiche, N., & Flach, P. (1999). 1BC: A first-order Bayesian classifier. In Proceedings of the 9th international workshop on inductive logic programming (pp. 92–103).
Lachiche, N., & Flach, P. (2003). Improving accuracy and cost of two-class and multi-class probabilistic classifiers using ROC curves. In Proceedings of the 20th international conference on machine learning (pp. 416–423).
Matsui, T. (1996). \(\mathsf{NP}\) -hardness of linear multiplicative programming and related problems. Journal of Global Optimization, 9, 113–119.
MOSEK ApS (2005). The MOSEK optimization tools version 3.2. http://www.mosek.com/.
Mossman, D. (1999). Three-way ROCs. Medical Decision Making, 19(1), 78–89.
O’Brien, D. B., & Gray, R. M. (2005). Improving classification performance by exploring the role of cost matrices in partitioning the estimated class probability space. In Proceedings of the ICML Workshop on ROC Analysis in Machine Learning (pp. 79–86).
Provost, F. J., & Fawcett, T. (1997). Analysis and visualization of classifier performance: Comparison under imprecise class and cost distributions. In Proceedings of the third international conference on knowledge discovery and data mining (KDD-97) (pp. 43–48).
Provost, F. J., & Fawcett, T. (1998). Robust classification systems for imprecise environments. In Proceedings of the 15th national conference on artificial intelligence (AAAI) (pp. 706–713).
Provost, F. J., & Fawcett, T. (2001). Robust classification for imprecise environments. Machine Learning, 42, 203–231.
Rockafellar, R. (1970). Convex analysis (2nd edn.). Princeton: Princeton University Press.
Schölkopf, B., & Smola, A. J. (2001). Learning with kernels. Cambridge: MIT Press.
Srinivasan, A. (1999). Note on the location of optimal classifiers in n-dimensional ROC space (Technical Report PRG-TR-2-99). Oxford University Computing Laboratory, Oxford.
Stoer, I. J., & Witzgall, C. (1996). Convexity and optimization in finite dimensions. Berlin: Springer.
Witten, I. H., et al. (2005). Weka machine learning toolbox. http://www.cs.waikato.ac.nz/ml/weka/.
Author information
Authors and Affiliations
Corresponding author
Additional information
Preliminary results appeared in Deng et al. (2006).
Editor: Tom Fawcett.
Rights and permissions
About this article
Cite this article
Bourke, C., Deng, K., Scott, S.D. et al. On reoptimizing multi-class classifiers. Mach Learn 71, 219–242 (2008). https://doi.org/10.1007/s10994-008-5056-8
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10994-008-5056-8