A Novel Manifold Regularized Online Semi-supervised Learning Algorithm
Ding, Shuguang; Xi, Xuanyang; Liu, Zhiyong; Qiao, Hong; Zhang, Bo; Liu, ZY
Conference Name23rd International Conference on Neural Information Processing (ICONIP)
Conference DateOCT 16-21, 2016
Conference PlaceKyoto, JAPAN
AbstractIn this paper, we propose a novel manifold regularized online semi-supervised learning ((OSL)-L-2) model in an Reproducing Kernel Hilbert Space (RK-HS). The proposed algorithm, named Model-BasedOnline Manifold Regularization (MOMR), is derived by solving a constrained optimization problem, which is different from the stochastic gradient algorithm used for solving the online version of the primal problem of Laplacian support vector machine (LapSVM). Taking advantage of the convex property of the proposed model, an exact solution can be obtained iteratively by solving its Lagrange dual problem. Furthermore, a buffering strategy is introduced to improve the computational efficiency of the algorithm. Finally, the proposed algorithm is experimentally shown to have a comparable performance to the standard batch manifold regularization algorithm.
KeywordManifold Regularization Online Semi-supervised Learning Lagrange Dual Problem
Indexed ByEI
Document Type会议论文
Corresponding AuthorLiu, ZY
Recommended Citation
GB/T 7714
Ding, Shuguang,Xi, Xuanyang,Liu, Zhiyong,et al. A Novel Manifold Regularized Online Semi-supervised Learning Algorithm[C],2016.
Files in This Item: Download All
File Name/Size DocType Version Access License
ICONIP.pdf(217KB)会议论文 开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Ding, Shuguang]'s Articles
[Xi, Xuanyang]'s Articles
[Liu, Zhiyong]'s Articles
Baidu academic
Similar articles in Baidu academic
[Ding, Shuguang]'s Articles
[Xi, Xuanyang]'s Articles
[Liu, Zhiyong]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Ding, Shuguang]'s Articles
[Xi, Xuanyang]'s Articles
[Liu, Zhiyong]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: ICONIP.pdf
Format: Adobe PDF
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.