DocumentCode :
46345
Title :
Max-Margin Discriminant Projection via Data Augmentation
Author :
Bo Chen ; Hao Zhang ; Xuefeng Zhang ; Wei Wen ; Hongwei Liu ; Jun Liu
Author_Institution :
Nat. Lab. of Radar Signal Process., Xidian Univ., Xi´an, China
Volume :
27
Issue :
7
fYear :
2015
fDate :
July 1 2015
Firstpage :
1964
Lastpage :
1976
Abstract :
In this paper, we introduce a new max-margin discriminant projection method, which takes advantage of the latent variable representation for support vector machine (SVM) as the classification criterion. Specifically, the proposed model jointly learns the discriminative subspace and classifier in a Bayesian framework by conditioning on augmented variables. Moreover, an extended nonlinear model is developed based on the kernel trick, where the similar model can be used in this setting with few modifications. To explore the sparsity in the kernel expansion, we use the spike-and-slab prior to seek basis vectors (BVs) from the corresponding candidates. Unlike existing methods, which employ BVs to approximate the original feature space, in our method BVs are sought to associate the final classification task. Thanks to the conditionally conjugate property, the parameters in our models can be inferred via the simple and efficient Gibbs sampler. Finally, we test our methods on synthesized and real-world data, including large-scale data sets to demonstrate their efficiency and effectiveness.
Keywords :
Markov processes; Monte Carlo methods; belief networks; pattern classification; support vector machines; BV; Bayesian framework; Gibbs sampler; SVM; basis vectors; classification criterion; conditionally conjugate property; data augmentation; discriminative subspace; extended nonlinear model; kernel expansion; latent variable representation; max-margin discriminant projection method; spike-and-slab prior; support vector machine; Bayes methods; Data models; Feature extraction; Hilbert space; Kernel; Support vector machines; Vectors; Feature extraction; Gibbs sampling; Kernel methods; Max-margin; Radar automatic target recognition (RATR); feature extraction; kernel methods; radar automatic target recognition (RATR);
fLanguage :
English
Journal_Title :
Knowledge and Data Engineering, IEEE Transactions on
Publisher :
ieee
ISSN :
1041-4347
Type :
jour
DOI :
10.1109/TKDE.2015.2397444
Filename :
7029124
Link To Document :
بازگشت