Abstract
In multi-view classification, the goal is to find a strategy for choosing the most consistent views for a given task. A strategy is a probability distribution over views. A strategy can be considered as advice given to an algorithm. There can be several strategies, each allocating a different probability mass to a view at different times. In this paper, we propose an algorithm for mining these strategies in such a way that its trust in a view for classification comes close to that of the best strategy. As a result, the most consistent views contribute to multi-view classification. Finally, we provide experimental results to demonstrate the effectiveness of the proposed algorithm.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
We chose MVK over SVM-2K because SVM-2K is inherently a two-view learning algorithm.
References
Brefeld, U., Buscher, C., Scheffer, T.: Multi-view hidden Markov perceptrons. In: Proceedings of GI Workshop
Busa-Fekete, R., Kegl, B.: Fast boosting using adversarial bandits. In: Proceedings of International Conference on Machine Learning (2010)
Crihalmeanu, S., Ross, A., Schukers, S., Hornak, L.: A protocol for multibiometric data acquisition, storage and dissemination. In: Technical report, WVU, Lane Department of Computer Science and Electrical Engineering (2007)
Cristianini, N., Shawe-Taylor, J.: An Introduction to Support Vector Machines and Other Kernel-based Learning Methods. Cambridge University Press, Cambridge (2000)
Dasgupta, S., Littman, M., McAllester, D.: PAC generalization bounds for co-training. In: Dietterich, T.G., Becker, S., Ghahramani, Z. (eds.) Advances in Neural Information Processing Systems, vol. 14, pp. 375–382. The MIT Press, Cambridge (2002)
Farquhar, J., Hardoon, D., Meng, H., Shawe-Taylor, J., Szedmak, S.: Two View Learning: SVM-2K Theory and Practice. Advances in Neural Information Processing Systems, vol. 18, pp. 355–362. MIT Press, Cambridge (2005)
Masek, L., Kovesi, P.: MATLAB source code for biometric identification system based on iris patterns. Technical report, The University of Western Australia (2003)
Minh, H., Bazzani, L., Murino, V.: A unifying framework for vector-valued manifold regularization and multi-view learning, pp. 100–108 (2013)
Schapire, R.E., Singer, Y.: Improved boosting algorithms using confidence rated predictions. Mach. Learn. 3(37), 297–336 (1999)
Wang, W., Zhou, Z.-H.: Analyzing co-training style algorithms. In: Kok, J.N., Koronacki, J., Lopez de Mantaras, R., Matwin, S., Mladenič, D., Skowron, A. (eds.) ECML 2007. LNCS (LNAI), vol. 4701, pp. 454–465. Springer, Heidelberg (2007)
Yu, S., Krishnapuram, B., Rosales, R., Rao, R.: Bayesian co-training. J. Mach. Learn. Res. 12, 2649–2680 (2011)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2016 Springer International Publishing Switzerland
About this paper
Cite this paper
Peng, J., Aved, A.J. (2016). Mining Best Strategy for Multi-view Classification. In: Tan, Y., Shi, Y. (eds) Data Mining and Big Data. DMBD 2016. Lecture Notes in Computer Science(), vol 9714. Springer, Cham. https://doi.org/10.1007/978-3-319-40973-3_27
Download citation
DOI: https://doi.org/10.1007/978-3-319-40973-3_27
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-40972-6
Online ISBN: 978-3-319-40973-3
eBook Packages: Computer ScienceComputer Science (R0)