| |
SVM is the most popular algorithm among kernel methods. Since SVM is formulated by quadratic programming, we can always obtain the optimal solution. Second-order cone programming (SOCP) is an extension of quadratic programming. We can solve SOCP efficiently, and also obtain the optimal solution. We have developed a new algorithm for multi-task learning using SOCP. We confirmed the usefulness of out algorithm in a variety of applications.
|
|
| |
[11] Tsuyoshi Kato, Hisashi Kashima, Masashi Sugiyama, Kiyoshi Asai:
Multi-task learning via conic programming,
J. C. Platt, D. Koller, Y. Singer, and S. Roweis (Eds.), Advances in Neural Information Processing Systems 20, pp.737-744, Cambridge, MA, MIT Press, 2008, presented at Twenty-First Annual Conference on Neural Information Processing Systems (NIPS2007)[pdf].
[12] Tsuyoshi Kato, Hisashi Kashima, Masashi Sugiyama, Kiyoshi Asai:
Conic Programming for Multi-Task Learning,
IEEE Transactions on Knowledge and Data Enginieering, Vol.22, No.7, pp. 957-968 (2010).
|
|
