Abstract.

This paper considers a problem of combining classification results from several observations of the same object. The task is seen as a case of collective decision making by a group of experts with estimated competence levels. Precision of different classification result combination methods is analyzed with different input data model, having per-frame character recognition results combination problem in video stream as an example. Experiments show that the strategy which selects a single most competent expert performs better with input data model without any non-relevant observations (in the context of character recognition in video stream — without characters location and segmentation errors). At the same time experiments show that strategies which combine several most competent experts using product rule or voting procedure outperform single-expect strategy with input data containing non-relevant observations.

Keywords:

decision theory, pattern recognition, recognition in video stream, ensemble classifiers.

PP. 45-55.

REFERENCES

1. Doermann D., Liang J., Huiping L. Proceedings of Seventh International Conference on Document Analysis and Recognition, Vol. 1, 2003, pp. 606-616.
2. V.V. Arlazarov, A. Zhukovsky, D. Krivtsov, D. Nikolaev, D. Polevoy. Analysis of using stationary and mobile small-scale digital cameras for documents recognition // Information technologies and computational systems, № 3, 2014, pp. 71-78.
3. K.B. Bulatov, D.A. Ilin, D.V. Polevoy, Yu.S. Chernyshova. Problems of machine-readable zones recognition using smallscale digital cameras of mobile devices // Proc. ISA RAS, Vol. 65/3, 2015, pp. 85-93.
4. T.S. Chernov, D.A. Ilin, P.V. Bezmaternykh, I.A. Faradjev, S.M. Karpenko. Research of methods for segmentation of document text block images using algorithms of structure analysis and machine learning]// Proc. RFBR. Image processing and pattern recognition, № 4 (92), 2016, pp. 55-71.
5. Rokach L. Ensemble-based classifiers. Artificial Intelligence Review, V. 33, Issue 1-2, 2010, pp. 1-39, doi:10.1007/s10462-009-9124-7.
6. Fumera G., Roli F. Linear combiners for classifier fusion: some theoretical and experimental results // Proceedings of the 4th
International Conference on Multiple Classifier Systems, 2003, pp. 74-83.
7. Schwenk H., Gauvain J.-L. Combining multiple speech recognizers using voting and language model information // IEEE International Conference on Speech and Language Processing, 2000, pp. 915-918.
8. Zhang C.-X., Duin R. An experimental study of one- and two- level classifier fusion for different sample sizes // Pattern Recognition Letters 32, 2011, pp. 1756-1767, doi:10.1016/j.patrec.2011.07.009.
9. Chen D. Text detection and recognition in images and video sequences. Thesis №2863, Lausanne, EPFL, 2003, 141 pages.
10. Wemhoener D., Yalniz I.Z., Manmatha R. Creating an improved version using noisy OCR from multiple editions // Proceedings of the 12th International Conference on Document Analysis and Recognition, 2013, pp. 160-164.
11. Lopresti D., Zhou J. Using consensus sequence voting to correct OCR errors // Computer Vision and Image Understanding,
V. 67(1), 1997, pp. 39-47.
12. Fiscus J.G. A post-processing system to yield reduced word error rates: Recognizer Output Voting Error Reduction (ROVER) // Proceedings of IEEE Workshop on Automatic Speech Recognition and Understanding, 1997, pp. 347-354, doi:10.1109/ASRU.1997.659110.
13. Kittler J. On combining classifiers // IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 20, № 3, 1998, pp. 226-239.
14. Rogova G. Combining the results of several neural network classifiers // Neural Networks, Vol. 7, № 5, pp. 777-781, 1994.
15. Quost B., Masson M.-H., Denoeux T. Classifier fusion in the Dempster-Shafer framework using optimized t-norm based combination rules // International Journal of Approximate Reasoning, Vol. 52, Issue 3, 2011, pp. 353-374.
16. Ting K.M., Witten I.H. Issues in stacked generalization // Journal of Artificial Intelligence Research, Vol. 10, 1999, pp. 271-289.
17. Kuncheva L.I., Bezdek J.C., Duin R.P.W. Decision templates for multiple classifier fusion: an experimental comparison // Pattern Recognition, Vol. 34, 2001, pp. 299-314.
18. Merz C. Using correspondence analysis to combine classifiers // Machine Learning, Vol. 36, 1999, pp. 33-58.
19. Nguyen T.T., Nguyen T.T.T., Pham X.C., Liew A.W.-C. A novel combining classifier method based on variational inference // Pattern Recognition, Vol. 49, 2016, pp. 198-212.
20. Ye P., Doermann D. Document image quality assessment: a brief survey // Proceedings of 12th International Conference on Document Analysis and Recognition, 2013, pp. 723-727.
21. Bulatov K., Polevoy D. Reducing overconfidence in neural networks by dynamic variation of recognizer relevance // Proceedings of 29th European Conference on Modelling and Simulation, 2015, pp. 488-491.
22. D.P. Nikolaev, D.V. Polevoy, T.S. Chernov. Method for automatic quality estimation of color segmentation within a problem of compressing images of printed documents // Trudy ISA RAN [Proc. ISA RAS], Vol. 63/3, 2013, pp. 78-84.
23. Arlazarov V.L., Emelyanov N.E. (Ed.) Documents flow. Applied aspects. Moscow, Editorial URSS, 2005, 184 p.
24. V.V. Arlazarov, K.B. Bulatov, S.M. Karpenko. Method for determining recognition confidence in the problem of embossed characters recognition // Proc. ISA RAS, Vol. 63/3, 2013, pp. 117-122.
25. Berend D., Kontorovich A. Consistency of weighted majority votes // Advances in Neural Information Processing Systems, Vol. 27, Issue 2, 2014, pp. 3446-3454.
26. Dzeroski S., Zenko B. Is combining classifiers better than selecting the best one? // Machine Learning, Vol. 45, Issue 3, 2004, pp. 255-273.
27. Ilin D., Krivtsov V. Creating training datasets for OCR in mobile video stream // Proceedings of 29th European Conference on Modelling and Simulation, 2015, pp. 516-520.
28. Nitzan S., Paroush J. Collective decision-making and jury theorems // The Oxford Handbook of Law and Economics: Volume 1: Methodology and Concepts, 2017, doi:10.1093/oxfordhb/9780199684267.013.035.