Abstract
A decomposition approach to multiclass classification problems consists in decomposing a multiclass problem into a set of binary ones. Decomposition splits the complete multiclass problem into a set of smaller classification problems involving only two classes (binary classification: dichotomies). With a decomposition, one has to define a recombination which recomposes the outputs of the dichotomizers in order to solve the original multiclass problem. There are several approaches to the decomposition, the most famous ones being one-against-all and one-against-one also called pairwise. In this paper, we focus on pairwise decomposition approach to multiclass classification with neural networks as the base learner for the dichotomies. We are primarily interested in the different possible ways to perform the so-called recombination (or decoding). We review standard methods used to decode the decomposition generated by a one-against-one approach. New decoding methods are proposed and compared to standard methods. A stacking decoding is also proposed which consists in replacing the whole decoding or a part of it by a trainable classifier to arbiter among the conflicting predictions of the pairwise classifiers. Proposed methods try to cope with the main problem while using pairwise decomposition: the use of irrelevant classifiers. Substantial gain is obtained on all datasets used in the experiments. Based on the above, we provide future research directions which consider the recombination problem as an ensemble method.
Similar content being viewed by others
References
Allwein E, Schapire R and Singer Y (2000). Reducing multiclass to binary: a unifying approach for margin classifiers. J Mach Learn Rese 1(2): 113–141
Bradley RA and Terry ME (1952). The rank analysis of incomplete block designs: I. The method of paired comparisons. Biometrika 39: 324–345
Campbell C (1997). Constructive learning techniques for designing neural network systems. Academic Press, San Diego
Cardot H, Lezoray O (2002) Graph of neural networks for pattern recognition. In: International Conference on Pattern Recognition (ICPR) vol 2. pp. 124–127.
Crammer K and Singer Y (2002). On the learnability and design of output codes for multiclass problems. Mach Learn 47(2–3): 201–233
Dietterich T and Bakiri G (1995). Solving Multiclass learning problems via error-coorecting output codes. J Artif Intelli Res 2: 263–286
Ding C and Dubchak I (2001). Multiclass protein fold recognition using support vector machines and neural networks. Bioinformatics 17: 349–358
Duan K, Keerthi S (2005) Which is the best Multiclass SVM method? An empirical study. In: International Workshop on Multiple Classifier Systems. pp 278–285
Frank E, Kramer S (2004) Ensembles of nested dichotomies for multi-class problems. In: ICML. pp 305–312
Friedman J (1996) Another approach to polychotomous classification. Technical report, Department of Statistics, Stanford University
Furnkranz J. (2002) Pairwise classification as an ensemble technique. In: European Conference on Machine Learning (ECML). pp 97–110
Furnkranz J (2002). Round Robin Classification. J Mach Learn Res 2: 721–747
Hastie T and Tibshirani R (1998). Classification by pairwise coupling. Ann Stat 26(2): 451–471
Hettich S, Blake C, Merz C (1998) UCI repository of machine learning databases. Technical report, University of California, Irvine, Department of Information and Computer Sciences
Hsu C-W and Lin C-J (2002). A comparison of methods for multi-class support vector machines. IEEE Transa Neural Netw 13(2): 415–425
Ie E, Weston J, Noble W, Leslie C (2005) Multi-class protein fold recognition using adaptive codes. In: International Conference on Machine Learning (ICML). pp 329–336
Klautau A, Jevtić N, Orlitsky A (2002) Combined binary classifiers with applications to speech recognition. In: International Conference on Spoken Language Processing (ICSLP). pp 2469–2472
Klautau A, Jevtić N and Orlitsky A (2003). On nearest neighbor error-correcting output codes with application to all-pairs multiclass support vector machnies. J Mach Learn Rese 4: 1–15
Knerr S, Personnaz L and Dreyfus G (1992). Handwritten digit recognition by neural networks with single-layer training. IEEE Trans Neural Netw 3(6): 962–968
Ko J, Kim E, Byun H (2004) Improved N-division output coding for multiclass learning problems. In: International Conference on Pattern Recognition (ICPR), vol. 3. pp 470– 473
Kressel U (1999) Advances in Kernel methods, support vector learning, Chapt. Pairwise classification and support vector machines. MIT Press
Kwok T-Y and Yeung D-Y (1997). Constructive algorithms for structure learning in feedforward neural networks for regression problems. IEEE Trans Neural Netw 8(3): 630–645
Lezoray O and Cardot H (2001). A neural network architecture for data classification. Int J Neural Syst 11(1): 33–42
Lezoray O, Cardot H (2005) Combining multiple pairwise neural networks classifiers: a comparative study. In: International Workshop on Artificial Neural Networks and Intelligent Information Processing (ANNIIP). pp 52–61
Lezoray O, Elmoataz A and Cardot H (2003). A color object recognition scheme: application to cellular sorting. Mach Vis Appl 14: 166–171
Lezoray O, Fournier D and Cardot H (2004). Neural network induction graph for pattern recognition. Neurocomputing C(57): 257–274
Lu B-L and Ito M (1999). Task decomposition and module combination based on class relations: a modular neural network for pattern classification. IEEE Trans Neural Netw 10(5): 1244–1256
Masulli F, Valentini G (2000) Comparing decomposition methods for classification. In: International Conference on Knowledge-Based Intelligent Engineering Systems and Allied Technologies, vol 2. pp 788–792
Mayoraz E, Alpaydin E (1999) Support vector machines for multi-class classification. In: International Work conference on Artificial Neural Networks, vol 2. pp 833–842
Moreira M, Mayoraz E (1998) Improved pairwise coupling classification with correcting classifiers. In: European Conference on Machine Learning (ECML). pp 160–171
Ou G, Murphey Y, Feldkamp A (2004) Multiclass pattern classification using neural networks. In: International Conference on Pattern Recognition (ICPR) vol. 4. pp 585–588
Phetkaew T, Kijsirikul B, Rivepiboon W (2003) Reordering adaptive directed acyclic graphs: an improved algorithm for multiclass support vector machines. In: International Joint Conference on Neural Networks (IJCNN), vol 2. pp 1605–1610
Platt J, Cristianini N, Shawe-Taylor J (2000) Large margin DAGs for multiclass classification. In: Advances in Neural Information Processing Systems (NIPS), vol 12. pp 547–553
Price D, Knerr S, Personnaz L (1995) Pairwise neural network classifiers with propabilistic outputs. In: Advances in Neural Information Processing Systems (NIPS), vol 7. pp 1109–116
Quinlan J (1993). C4.5: programs for machine learning. Morgann Kauffman, San Mateo
Rifkin R and Klautau A (2004). In defense of one-vs-all classification. J Mach Learn Rese 5: 101–141
Savicky P, Furnkranz J (2003) Combining pairwise classifiers with stacking. In: Intellignent Data Analysis (IDA). pp 219–229
Takahashi F, Abe S (2003) Optimizing directed acyclic graph support vector machines. In: Artificial Neural Networks in Pattern Recognition (ANNPR). pp 166–173
Tax D, Duin R (2002) Using two-class classifiers for multiclass classification. In: International Conference on Pattern Recognition (ICPR), vol 2. pp 124–127
Vural V, Dy JG (2004) A hierarchical method for multi-class support vector machines. In: International Conference on Machine Learning (ICML). p 105
Wolpert D (1992). Stacked generalization. Neural Netw 5(2): 241–260
Zhou Z-H and Jiang Y (2003). NeC.45: neural ensemble based C.45. IEEE Trans Knowl Data Eng 16(6): 770–773
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Lézoray, O., Cardot, H. Comparing Combination Rules of Pairwise Neural Networks Classifiers. Neural Process Lett 27, 43–56 (2008). https://doi.org/10.1007/s11063-007-9058-5
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11063-007-9058-5