zbMATH — the first resource for mathematics

Methods and algorithms of collective recognition. (English. Russian original) Zbl 1163.93307
Autom. Remote Control 69, No. 11, 1821-1851 (2008); translation from Avtom. Telemekh. 2008, No. 11, 3-40 (2008).
Summary: The collective decision making and, in particular, the collective recognition is treated as the problem of joint application of multiple classifier decisions. The decisions are made about the class of an entity, situation, image, etc. The joint decision is used to improve quality of the final decision by aggregation and coordination of different classifier decisions using a metalevel algorithm. The studies in the field of collective recognition, which were started in the middle of the 1950s, find wide application in practice during the last decade. Since they are used for solving complex large-scale applied problems, the interest of both theoretical scientists and engineers is focused on them. A new impetus for the studies was given by the recent development of embedded distributed structures involving ensembles of intellectual sensors that make decisions under uncertainties on the base of limited local information. The final decision of high quality, in particular, the decision of higher aggregation level, is made by combining local classifier decisions on the metalevel. There are dozens of recent publications proposing new ideas and new approaches and algorithms of collective recognition. Unfortunately, some papers rediscover results published several decades before. The goal of this review is to present the main ideas of collective recognition and to outline the status of researches basing on the original source works. The review covers the period from the 1950s, when the first ideas and methods appeared, up to present time.
93A15 Large-scale systems
62H30 Classification and discrimination; cluster analysis (statistical aspects)
C4.5; UCI-ml
Full Text: DOI
[1] Luce, D.R. and Raiffa, H., Games and Decisions, New York: Wiley, 1957. · Zbl 0084.15704
[2] Mirkin, B.G., Problema gruppovogo vybora (Problem of Group Choice), Moscow: Nauka, 1974. · Zbl 0302.62029
[3] Blin, J., Fu, K., and Whinston, A., Application of Pattern Recognition to some Problems in Economics, Techniq. Optim., 1972, no. 416, pp. 1–18.
[4] Kanal, L., Interactive Pattern Analysis and Classification. Survey and Commentary, Proc. IEEE, 1972, no. 10, pp. 1200–1215.
[5] Matematicheskie metody v sotsial’nykh naukakh (MathematicalMathods in Social Sciences), Lazarfeld, P. and Henry, N., Eds., Moscow: Nauka, 1973.
[6] Beshelev, S.D. and Gurvitch, F.G., Ekspertnye otsenki (Expert Judgements), Moscow: Nauka, 1973.
[7] Vorobiev, N.N., Mathematization Problems of Decision Making Basing on Expert Judgements, Proc. IV Symp. Cybernatics, 1972, vol. 3, pp. 47–51.
[8] Glushkov, V.M., On Predictions Based on Expert Judgements, Kibernetika, 1969, no. 2, pp. 2–4.
[9] von Neumann, J., Probabilistic Logics and Synthesis of Reliable Organisms from Unreliable Components, in Automata Studies, Shannon, C.E. and McCarthy, J., Eds., Princeton: Princeton Univ. Press, 1956, pp. 43–98.
[10] Rastrigin, L.A. and Erenshtein, R.Kh., Metod kollectivnogo raspoznavaniya (Collective Recognition Method), Moscow: Energoizdat, 1981.
[11] Condorcet, N.C., Essai sur l’application de l’analyse à la probabilité des decisions rendues a la pluralité des voi, Paris: Imprimerie Royale, 1785.
[12] Rastrigin, L.A. and Erenshtein, R.Kh., Collective of Algorithms of Decision Rules in Pattern Recognition Problems, Izv. AN SSSR, Tekh. Kibern., 1978, no. 2, pp. 116–126.
[13] Rastrigin, L.A. and Erenshtein, R.Kh., Collective of Algorithms of Decision Rules in Pattern Recognition Problems, Avtom. Telemekh., 1975, no. 9, pp. 134–144.
[14] Rastrigin, L.A. and Erenshtein, R.Kh., A Collective of Algorithms, Proc. of the 4th IJCAI, Tbilisi, 1975, pp. 370–373.
[15] Rastrigin, L.A. and Erenshtein, R.Kh., Learning of a Collective of Solving Rules, in Adaptive Systems, Riga: Zinatne, 1974, vol. 4, pp. 8–20.
[16] Kittler, J., Hatef, M., Duin, R.P.W., and Matas, J., On Combining Classifiers, IEEE Trans. Pattern Anal. Machine Intelligence, 1998, vol. 20, no. 3, pp. 226–239. · doi:10.1109/34.667881
[17] Kuncheva, L., Bezdec, J., and Duin, R.P.W., Decision Templates for Multiple Classifier Fusion, Pattern Recogn., 2001, vol. 34, no. 2, pp. 299–314. · Zbl 0991.68064 · doi:10.1016/S0031-3203(99)00223-X
[18] Ho, T.K., Multiple Classifier Combination: Lessons and Next Steps, in Hybrid Methods in Pattern Recognition, Kandel, A. and Bunke, H., Eds., Singapore: World Scientific, 2002, pp. 171–198. · Zbl 1012.68167
[19] Datta, S., Bhaduri, K., Giannella, C., Wolff, R., and Kargupta, H., Distributed Data Mining in Peer-to-Peer Networks, IEEE Internet Computing, Special issue on Distributed Data Mining, 2006, vol. 10, no. 4, pp. 18–26.
[20] Fix, E. and Hodges, J., Nonparametric Discrimination, Texas: USAF School of Aviation Medicine, 1951.
[21] Rosenblatt, F., Principles of Neurodynamics: Perceptrons and the Theory of Brain Mechanisms, Washington: Spartan Books, 1962. · Zbl 0143.43504
[22] Bongard, M.M., Problema uznavaniya (Recognition Problem), Moscow: Nauka, 1967.
[23] Clark, P. and Niblett, T., The CN2 Induction Algorithm, Machine Learning J., 1989, no. 3, pp. 261–283.
[24] Gorodetski, V. and Karsayev, O., Algorithm of Rule Extraction from Learning Data, Proc. Int. Conf. Expert Syst. & Artificial Intelligence (EXPERSYS-96), 1996, pp. 133–138.
[25] Michalski, R.A., Theory and Methodology of Inductive Learning, Machine Learning, 1983, no. 1, pp. 83–134.
[26] Ali, K. and Pazzani, M., Error Reduction through Learning Multiple Descriptions, Machine Learning, 1996, vol. 24, no. 3, pp. 173–202.
[27] Clemen, R., Combining Forecasts: A Review and Annotated Bibliography, Int. J. Forecast, 1989, no. 5, pp. 559–583.
[28] Dietterich, T., Machine Learning Research: Four Current Directions, AI Magazine, 1997, vol. 18, no. 4, pp. 97–136.
[29] Dietterich, T., Ensemble Methods in Machine Learning, LNCS, 2000, no. 1857, pp. 1–15. · Zbl 1073.68712
[30] Buntine, W.L., A Theory of Learning Classification Rules, Sydney: Univ. of Technology, 1990.
[31] Hashem, S., Optimal Linear Combination of Neural Networks, Purdue: School of Industrial Engineering, 1997.
[32] Jordan, M. and Jacobs, R., Hierarchical Mixtures of Experts and the EM Algorithm, Neural Comput., 1994, vol. 6, no. 2, pp. 181–214. · doi:10.1162/neco.1994.6.2.181
[33] Perrone, M. and Cooper, L., When Networks Disagree: Ensemble Methods for Hybrid Neural Networks, in Neural Networks for Speech Image Proc., 1993, pp. 126–142.
[34] Breiman, L., Bagging Predictors, Machine Learning, 1996, vol. 24, no. 2, pp. 123–140. · Zbl 0858.68080
[35] Freund, Y. and Shapire, R., Experiments with a New Boosting Algorithm, Proc. 13th Int. Conf. Machine Learning, 1996, pp. 148–156.
[36] Moerland, P., Mixtures of Experts Estimate A Posteriori Probabilities, Proc. Int. Conf. Artificial Neural Networks (ICANN’97), 1997, pp. 499–504.
[37] Ortega, J., Coppel, M., and Argamon, S., Arbilearning Among Competing Classifiers Using Learned Referees, Knowledge Inform. Syst., 2001, no. 3, pp. 470–490. · Zbl 0987.68629
[38] Gorodetskiy, V., Karsaev, O., Samoilov, V., and Serebryakov, S., P2P Agent Platform: Implementation and Testing, AAMAS Sixth Int. Workshop Agents Peer-to-Peer Comput., 2007, pp. 21–32.
[39] Ting, K. and Witten, I., Issues in Stacked Generalization, J. Artific. Intellig. Res., 1999, no. 10, pp. 271–289. · Zbl 0915.68075
[40] Wolpert, D., Stacked Generalization, Neural Network, 1992, vol. 5, no. 2, pp. 241–260. · doi:10.1016/S0893-6080(05)80023-1
[41] Breiman, L., Stacked Regression, Machine Learning, 1996, vol. 24, no. 1, pp. 49–64. · Zbl 0849.68104
[42] Stolfo, S. and Chan, P., A Comparative Evaluation of Voting and Meta-Learning on Partitioned Data, Proc. Int. Conf. Machine Learning, 1995, pp. 90–98.
[43] Merz, C. and Murphy, P., UCI Repository on Machine Learning Databases, Irvine: Univ. of California, 1997 ( http://www.ics.uci.edu/mlearn/MLRrepository.html (21.04.06)).
[44] Merz, C., Using Correspondence Analysis to Combining Classifiers, Machine Learning, 1999, no. 36, pp. 33–58.
[45] Bay, S.D. and Pazzani, M.J., Characterizing Model Error and Differences, Proc. 17 Int. Conf. Machine Learning (ICML-2000), 2000, pp. 49–56.
[46] Quinlan, R., C4.5 Programs for Machine Learning, San Francisco: Morgan Kaufmann, 1993.
[47] Murthy, S., Kassif, S., Salzberg, S., and Beigel, R., OC1: Randomized Induction of Oblique Decision Trees, Proc. AAAI-93, 1993, pp. 322–327.
[48] Cost, S. and Salzberg, S., A Weighted Nearest Neighbor Algorithm for Learning with Symbolic Features, Machine Learning, 1993, vol. 10, no. 1, pp. 57–78.
[49] Rumelhart, D.E, Hinton, G.E., and Williams, R.J., Learning Internal Representation by Error Propagation, Parallel Distributed Proc. Exploration Microstructure Cognit., 1986, no. 1, pp. 318–362.
[50] Gama, J. and Brazdil, P., Cascade Generalization, Machine Learning, 2000, vol. 41, no. 3, pp. 315–342. · Zbl 0966.68180 · doi:10.1023/A:1007652114878
[51] Niyogi, P., Pierrot, J.-B., and Siohan, O., Multiple Classifiers by Constrained Minimization, Proc. Int. Conf. Acoustics, Speech, Signal Proc., 2000, no. 6. pp. 3462–3465.
[52] Prodomidis, A., Chan, P., and Stolfo, S., Meta-Learning in Distributed Data Mining Systems: Issues and Approaches, in Advances in Distributed Data Mining, Kargupta, H. and Chan, P., Eds., Book AAAI Press, 2000.
[53] Ting, K., The Characterization of Predictive Accuracy and Decision Combination, Proc. 13 Int. Conf. Machine Learning, 1996, pp. 498–506.
[54] Amanda, J.C. and Sharkey, N.E., Combining Artificial Neural Nets: Ensembles and Modular Multi–Net Systems, New York: Springer-Verlag, 1999. · Zbl 0910.00025
[55] Seewald, A. and Fuernkranz, J., An Evaluation of Grading Classifiers, Proc. 4 Int. Conf. Intelligent Data Anal., 2001, pp. 115–124. · Zbl 1029.68895
[56] Todorovski, L. and Dzeroski, S., Combining Classifiers with Meta Decision Trees, Proc. 4 Eur. Conf. Principles of Data Mining and Knowledge Discovery (PKDD-2000), 2000, pp. 54–64.
[57] Todorovski, L. and Dzeroski, S., Combining Classifiers with Meta Decision Trees, Machine Learning, 2003, vol. 50, no. 3, pp. 223–249. · Zbl 1033.68099 · doi:10.1023/A:1021709817809
[58] Huang, T., Hess, C., Pan, H., and Liang Zhi-Pei, A Neuronet Approach to Information Fusion, Proc. IEEE First Workshop Multimedia Signal Proc., 1997, pp. 45–50.
[59] Dar-Shyang, L., A Theory of Classifier Combination: The Neural Network Approach, Proc. Int. Conf. Document Anal. and Recognit. (ICDAR), 1995, pp. 42–45.
[60] Komartsova, L.G. and Maksimov, A.V., Neirokompyutery (Neurocomputers), Moscow: Bauman State Univ. of Technology, 2002, pp. 115–117.
[61] Lipnickas, A. and Korbicz, J., Adaptive Selection of Neural Networks for a Committee Decision, IEEE Int. Workshop Intelligent Data Acquisition and Advanced Comput. Syst.: Technology and Appl., 2003, pp. 109–114.
[62] McKay, T., Classifier Ensembles: A Practical Overview ( http://www.music.mcgill.ca/:_cmckay/software/computer_science/Classifier Ensembles/greeting_page.html (26.05.08)).
[63] Jacobs, R., Jordan, M., Nowlan, S., and Hinton, G., Adaptive Mixtures of Local Experts, Neural Comput., 1991, no. 3, pp. 79–87.
[64] Waterhouse, S. and Robinson, A., Classification Using Hierarchical Mixtures of Experts, Proc. IEEE Workshop Neural Networks Signal Proc., 1994, pp. 177–186.
[65] Kuncheva, L., Switching between Selection and Fusion in Combining Classifiers: An Experiment, IEEE Trans. Syst. Man Cybernet., 2002, no. 32, pp. 146–156.
[66] Kuncheva, L. and Whitaker, C., Measures of Diversity in Classifier Ensembles, Machine Learning, 2003, no. 51, pp. 181–207. · Zbl 1027.68113
[67] Kleinberg, E.M., Stochastic Discrimination, Ann. Mathemat. Artific. Intellig., 1990, no. 1, pp. 207–239. · Zbl 0870.68071
[68] Kleinberg, E.M., A Mathematically Rigorous Foundation for Supervised Learning, Lecture Notes Comput. Sci., 2000, no. 1857, pp. 67–78.
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.