Skip to Main content Skip to Navigation
Conference papers

MKPM: A multiclass extension to the kernel projection machine

Abstract : We introduce Multiclass Kernel Projection Machines (MKPM), a new formalism that extends the Kernel Projection Machine framework to the multiclass case. Our formulation is based on the use of output codes and it implements a co-regularization scheme by simultaneously constraining the projection dimensions associated with the individual predictors that constitute the global classifier. In order to solve the optimization problem posed by our formulation, we propose an efficient dynamic programming approach. Numerical simulations conducted on a few pattern recognition problems illustrate the soundness of our approach.
Complete list of metadatas

Cited literature [18 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-00669032
Contributor : Liva Ralaivola <>
Submitted on : Friday, February 10, 2012 - 11:57:06 PM
Last modification on : Thursday, January 23, 2020 - 6:22:14 PM
Document(s) archivé(s) le : Wednesday, December 14, 2016 - 4:51:27 AM

File

multiclasskpm.pdf
Files produced by the author(s)

Identifiers

Collections

Citation

Sylvain Takerkart, Liva Ralaivola. MKPM: A multiclass extension to the kernel projection machine. IEEE Computer Vision and Pattern Recognition, Jun 2011, Colorado Springs, France. pp.2785-2791, ⟨10.1109/CVPR.2011.5995657⟩. ⟨hal-00669032⟩

Share

Metrics

Record views

551

Files downloads

281