The Distance Weighted Discrimination Method was proposed in Marron et al (2007), which improves the performance of SVM method when the number of variables is close to or larger than the number of observaations. There are several important extensions of the DWD method, which try to improve the performance interms of interpretability and unbalance data sets, e.g. sparse DWD (Zhang and Lin, 2008) and weighted DWD (Qiao et al, 2008). In this website, I provide some links to this work and some matlab routines designed by them or myself for testing. If you find any problems for these matlab functions, please share your feedback with me. Thank you very much!
1. Regular DWD method
Marron, J. S., Todd, M. J. and Ahn, J. (2007) Distance-Weighted Discrimination, Journal of the American Statistical Association, Vol. 102, No. 480, pp. 1267-1271.
Marron's code for regular DWD method, Ahn's website for the DWD method, Lingsong's experimental code for the DWD method (needs the SDPT3 package preinstalled)
2. Weighted DWD method
Qiao, X., Zhang, H. H., Liu, Y., Todd, M. J. and Marron, J. S.(2008), Asymptotic Properties of Distance-Weighted Discrimination, submitted.
Lingsong's experimental code for the weighted DWD method (needs the SDPT3 pacakge preinstalled)
3. Sparse DWD method
Zhang, L. and Lin, X. (2008), Sparse Distance Weighted Discrimination Method, submitted.
Lingsong's code for SDWD methods, which allows L1-DWD and adaptive L1-DWD methods. (needs the above experimental code of the DWD method, and the SDPT3 packages)
4. Sparse Weighted DWD method
Collaboration work with Qiao, Liu and Marron, and possible other collaborators.
Lingsong's code for SWDWD method, which allows L1-wDWD and adaptive L1-wDWD methods. (needs the above experimental code of the wDWD method, and the SDPT3 pacakage)
5. SDPT package
This package is designed by Kim-Chuan Toh , Michael J. Todd, and Reha H. Tutuncu to solve semi-defimite programming problems.
Download the package from Toh's webstie; local download the version 4 (beta)
This website is updated on July 21, 2009
