Abstract
We revise Nesterov’s Accelerated Gradient (NAG) procedure for the SVM dual problem and propose a strictly monotone version of NAG that is capable of accelerating the second order version of the SMO algorithm. The higher computational cost of the resulting Nesterov Accelerated SMO (NA–SMO) is twice as high as that of SMO so the reduction in the number of iterations is not likely to translate in time savings for most problems. However, understanding NAG is presently an area of strong research and some of the resulting ideas may offer venues for even faster versions of NA–SMO.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Allen-Zhu, Z., Orecchia, L.: Linear coupling: an ultimate unification of gradient and mirror descent. arXiv:1407.1537 (2014)
Arjevani, Y., Shalev-Shwartz, S., Shamir, O.: On lower and upper bounds for smooth and strongly convex optimization problems. CoRR abs/1503.06833 (2015)
Beck, A., Teboulle, M.: Fast gradient-based algorithms for constrained total variation image denoising and deblurring problems. IEEE Trans. Image Process. 18(11), 2419–2434 (2009)
Beck, A., Teboulle, M.: A fast iterative shrinkage-thresholding algorithm for linear inverse problems. SIAM J. Imaging Sci. 2(1), 183–202 (2009)
Bubeck, S., Lee, Y.T., Singh, M.: A geometric alternative to Nesterov’s accelerated gradient descent. arXiv:1506.08187 (2015)
Chambolle, A., Dossal, C.: How to make sure the iterates of FISTA converge. https://hal.inria.fr/hal-01060130
Chang, C.C., Lin, C.J.: LIBSVM: a library for support vector machines. ACM Trans. Intell. Syst. Technol. 2(3), 27:1–27:27 (2011)
Chen, P.H., Fan, R.E., Lin, C.J.: A study on SMO-type decomposition methods for support vector machines. IEEE Trans. Neural Networks 17, 893–908 (2006)
Flammarion, N., Bach, F.R.: From averaging to acceleration, there is only a step-size. In: Proceedings of the 28th Conference on Learning Theory, COLT 2015, Paris, France, 3–6 July 2015, pp. 658–695 (2015)
Nesterov, Y.: Introductory Lectures on Convex Optimization: A Basic Course. Applied Optimization. Kluwer Academic Publishers, Boston (2004)
O’Donoghue, B., Candès, E.J.: Adaptive restart for accelerated gradient schemes. Found. Comput. Math. 15(3), 715–732 (2015)
Su, W., Boyd, S., Candes, E.: A differential equation for modeling nesterovs accelerated gradient method: theory and insights. In: Ghahramani, Z., Welling, M., Cortes, C., Lawrence, N.D., Weinberger, K.Q. (eds.) Advances in Neural Information Processing Systems, vol. 27, pp. 2510–2518 (2014)
Sutskever, I., Martens, J., Dahl, G.E., Hinton, G.E.: On the importance of initialization and momentum in deep learning. In: Dasgupta, S., Mcallester, D. (eds.) Proceedings of the 30th International Conference on Machine Learning (ICML 2013), vol. 28, pp. 1139–1147 (2013)
Wibisono, A., Wilson, A., Jordan, M.: A variational perspective on accelerated methods in optimization. arXiv:1603.04245 (2016)
Acknowledgments
With partial support from Spain’s grants TIN2013-42351-P and S2013/ICE-2845 CASI-CAM-CM, and also of the Cátedra UAM–ADIC in Data Science and Machine Learning. The first author is also supported by the FPU–MEC grant AP-2012-5163. The authors also gratefully acknowledge the use of the facilities of Centro de Computación Científica (CCC) at UAM.
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
Torres-Barrán, A., Dorronsoro, J.R. (2016). Nesterov Acceleration for the SMO Algorithm. In: Villa, A., Masulli, P., Pons Rivero, A. (eds) Artificial Neural Networks and Machine Learning – ICANN 2016. ICANN 2016. Lecture Notes in Computer Science(), vol 9887. Springer, Cham. https://doi.org/10.1007/978-3-319-44781-0_29
Download citation
DOI: https://doi.org/10.1007/978-3-319-44781-0_29
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-44780-3
Online ISBN: 978-3-319-44781-0
eBook Packages: Computer ScienceComputer Science (R0)