Hyperparameter Selection for APC-ECOC

다국어 초록 (Multilingual Abstract)

The main object of this paper is to develop a leave-one-out(LOO) bound of all pairwise comparison error correcting output codes (APC-ECOC). To avoid using classifiers whose corresponding target values are 0 in APC-ECOC and requiring pilot estimates we developed a bound based on mean misclassification probability(MMP). It can be used to tune kernel hyperparameters. Our empirical experiment using kernel mean squared estimate(KMSE) as the binary classifier indicates that the bound leads to good estimates of kernel hyperparameters.

참고문헌 (Reference)

1 Schapire,R.E, "Using output codes to boost multi-class learning problems" Morgan Kaufman 313-321, 1997

2 Kuncheva,L.I, "Using diversity measures for generating error -correcting output codes in classifier ensembles" 26 (26): 83-90, 2005

3 Vapnik,V, "Statistical Learning Theory" Springer, New York 1998

4 Seok, K. H, "Some comments on error correcting output codes" 4223 : 2006-, 2006

5 Dietterich, T. G, "Solving multi-class learning problems via error-correcting output codes" 2 : 263-286, 1995

6 Allwein, E. L, "Reducing multi-class to binary : A unifying approach for margin classifiers" 1 : 113-141, 2000

7 Tapia, E, "On the design of error adaptive ECOC algorithms" 2001

8 Passerini, A, "New results on error correcting output codes of kernel machines" 15 (15): 45-54, 2004

9 Lee, Y, "Multicategory support vector machines: theory and application to the classification of microarray data and satellite radiance data" 99 (99): 67-81, 2004

10 Kuncheva, L. I, "Measures of diversity in classifier ensembles" 51 : 181-207, 2003

11 Elisseeff, A, "Leave-one-out error and stability of learning algorithms with applications. In Advances in Learning Theory: Methods, Models and Applications" IOS press 190 : 2003

12 Xu, J, "Kernel MSE algorithm: A unified framework for KFD, LS-SVM" 2 : 1486-1491, 2001

13 Kittler, J, "Face verification using error correcting output codes. In Computer Vision and Pattern Recognition CVPR01" Hawaii, IEEE Press 2001

14 Jaakkola, T, "Exploiting generative modes in discriminative classifiers" 1998 1998

15 Berger,A, "Error-correcting output coding for text classification" 1999

16 Dietterich, T. G, "Error-correcting output codes : A general method for improving multi-class inductive learning programs" AAAI Press 572-577, 1991

17 Pujol, O, "Discriminant ECOC: A heuristic method for application dependent design of error correcting output codes" 28 (28): 1007-1012, 2006

18 석경하, "Data-Adaptive ECOC for Multicategory Classification" 한국데이터정보과학회 19 (19): 25-36, 2008

19 석경하, "Comparison of Various Criteria for Designing ECOC" 한국데이터정보과학회 17 (17): 2-, 2006

20 Windeatt, T, "Coding and decoding strategies for multi-class learning problems" 4 (4): 11-21, 2003

21 Aha, D. W, "Cloud classification using error correcting output codes. Artificial Intelligence Applications : Natural Resources" 11 (11): 13-28, 1997

22 Hastie, T, "Classification by pairwise coupling" 26 : 451-471, 1998

23 Breiman,L, "Bagging predictors" 24 : 3-140, 1997

24 Masulli, F, "An experimental analysis of the dependence among codeword bit error in ECOC learning machines" 57 : 189-214, 2004

25 Zhang,T, "A leave one out cross validation bound for kernel methods with application in learning" 43 : 427-433, 2001

26 Freund, Y, "A decision-theoretic generalization of on-line learning and application to boosting" 55 : 119-139, 1997