Abstract
Traditional supervised approaches realize an inductive learning process: A model is learnt from labeled examples, in order to predict the labels of unseen examples. On the other hand, transductive learning is less ambitious. It can be thought as a procedure to learn the labels on a training set, while, simultaneously, trying to guess the best labels on the test set. Intuitively, transductive learning has the advantage of being able to directly use training patterns while deciding on a test pattern. Thus, transductive learning faces a simpler problem with respect to inductive learning. In this paper, we propose a preliminary comparative study between a simple transductive model and a pure inductive model, where the learning architectures are based on feedforward neural networks. The goal is to understand how transductive learning affects the complexity (measured by the number of hidden neurons) of the exploited neural networks. Preliminary experimental results are reported on the classical two spirals problem.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
Actually, the main goal of transductive learning, as proposed in the present work, is to diffuse information coming from neighbor data to improve the whole classification accuracy. Technically speaking, we face a fully supervised problem, defining first the concept of data vicinity, and then training a feedforward neural network also on the base of the target information on the neighbors. Such a simplification is required in order to compare learning by induction and learning by transduction.
- 2.
Notice that even if several prototypes are used for each pattern, a single network \(N_\mathbf{w}\) is trained.
- 3.
It can be easily shown that the required number of hiddens increases with the length of the spirals and with the noise in the generation of the patterns.
- 4.
Linear–output classifiers are experimentally proved to work well in many practical problems, especially for high dimensional input spaces, reaching accuracy levels comparable to non–linear classifiers while taking less time to be trained and used [19]. Moreover, they are not affected by the saturation problems, which can arise in sigmoid neurons.
References
Vapnik, V.: The Nature Of Statistical Learning Theory. Springer Science & Business Media, New York (2013)
Belkin, M., Niyogi, P., Sindhwani, V.: Manifold regularization: a geometric framework for learning from labeled and unlabeled examples. J. Mach. Learn. Res. 7, 2399–2434 (2006)
Zhou, D., Bousquet, O., Lal, T.N., Weston, J., Schölkopf, B.: Learning with local and global consistency. Adv. Neural Inf. Process. Syst. 16, 321–328 (2004)
Zhu, X., Ghahramani, Z., Lafferty, J., et al.: Semi-supervised learning using Gaussian fields and harmonic functions. ICML 3, 912–919 (2003)
Blum, A., Mitchell, T.: Combining labeled and unlabeled data with co-training. In: Proceedings of the 11th Annual Conference on Computational Learning Theory, pp. 92–100. ACM (1998)
El-Yaniv, R., Pechyony, D., Vapnik, V.: Large margin vs. large volume in transductive learning. Mach. Learn. 72, 173–188 (2008)
Ifrim, G., Weikum, G.: Transductive learning for text classification using explicit knowledge models. In: Fürnkranz, J., Scheffer, T., Spiliopoulou, M. (eds.) PKDD 2006. LNCS (LNAI), vol. 4213, pp. 223–234. Springer, Heidelberg (2006). doi:10.1007/11871637_24
Nigam, K., McCallum, A.K., Thrun, S., Mitchell, T.: Text classification from labeled and unlabeled documents using EM. Mach. Learn. 39, 103–134 (2000)
Blake, A., Rother, C., Brown, M., Perez, P., Torr, P.: Interactive image segmentation using an adaptive GMMRF model. In: Pajdla, T., Matas, J. (eds.) ECCV 2004. LNCS, vol. 3021, pp. 428–441. Springer, Heidelberg (2004). doi:10.1007/978-3-540-24670-1_33
Balcan, M., Blum, A., Choi, P., Lafferty, J., Pantano, B., Rwebangira, M., Zhu, X.: Person identification in webcam images: an application of semi-supervised learning. In: Proceedings of the 22nd International Conference on Machine Learning (ICML05), Workshop on Learning with Partially Classified Training Data, pp. 1–9 (2005)
Duh, K., Kirchhoff, K.: Lexicon acquisition for dialectal Arabic using transductive learning. In: Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing, pp. 399–407. Association for Computational Linguistics (2006)
Ueffing, N., Haffari, G., Sarkar, A., et al.: Transductive learning for statistical machine translation. In: Annual Meeting-Association for Computational Linguistics, vol. 45, p. 25 (2007)
Lane, T.: A decision-theoritic, semi-supervised model for intrusion detection. In: Malo, M.A. (ed.) Machine Learning and Data Mining for Computer Security, pp. 157–177. Springer, Heidelberg (2006)
Vert, J.P., Yamanishi, Y.: Supervised graph inference. In: Advances in Neural Information Processing Systems, pp. 1433–1440 (2004)
Craig, R.A., Liao, L.: Transductive learning with EM algorithm to classify proteins based on phylogenetic profiles. Int. J. Data Min. Bioinf. 1, 337–351 (2007)
Weston, J., Pérez-Cruz, F., Bousquet, O., Chapelle, O., Elisseeff, A., Schölkopf, B.: Feature selection and transduction for prediction of molecular bioactivity for drug design. Bioinformatics 19, 764–771 (2003)
Bair, E., Tibshirani, R.: Semi-supervised methods to predict patient survival from gene expression data. PLoS Biol. 2, e108 (2004)
Hughes, N.P., Roberts, S.J., Tarassenko, L.: Semi-supervised learning of probabilistic models for ECG segmentation. In: 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, IEMBS 2004, vol. 1, pp. 434–437. IEEE (2004)
Yuan, G.X., Ho, C.H., Lin, C.J.: Recent advances of large-scale linear classification. Proc. IEEE 100(9), 2584–2603 (2012)
Bianchini, M., Scarselli, F.: On the complexity of neural network classifiers: a comparison between shallow and deep architectures. IEEE Trans. Neural Netw. Learn. Syst. 25, 1553–1565 (2014)
Kurková, V., Sanguineti, M.: Model complexities of shallow networks representing highly varying functions. Neurocomputing 171, 598–604 (2016)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2016 Springer International Publishing AG
About this paper
Cite this paper
Bianchini, M., Belahcen, A., Scarselli, F. (2016). A Comparative Study of Inductive and Transductive Learning with Feedforward Neural Networks. In: Adorni, G., Cagnoni, S., Gori, M., Maratea, M. (eds) AI*IA 2016 Advances in Artificial Intelligence. AI*IA 2016. Lecture Notes in Computer Science(), vol 10037. Springer, Cham. https://doi.org/10.1007/978-3-319-49130-1_21
Download citation
DOI: https://doi.org/10.1007/978-3-319-49130-1_21
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-49129-5
Online ISBN: 978-3-319-49130-1
eBook Packages: Computer ScienceComputer Science (R0)