Robust C-loss kernel classifiers

Robust C-loss kernel classifiers. Xu, G., Hu, B., & Principe, J. C. IEEE Transactions on Neural Networks and Learning Systems, 29(3):510–522, March, 2018.

Paper doi abstract bibtex

The correntropy-induced loss (C-loss) function has the nice property of being robust to outliers. In this paper, we study the C-loss kernel classiﬁer with the Tikhonov regularization term, which is used to avoid overﬁtting. After using the halfquadratic optimization algorithm, which converges much faster than the gradient optimization algorithm, we ﬁnd out that the resulting C-loss kernel classiﬁer is equivalent to an iterative weighted least square support vector machine (LS-SVM). This relationship helps explain the robustness of iterative weighted LS-SVM from the correntropy and density estimation perspectives. On the large-scale data sets which have low-rank Gram matrices, we suggest to use incomplete Cholesky decomposition to speed up the training process. Moreover, we use the representer theorem to improve the sparseness of the resulting C-loss kernel classiﬁer. Experimental results conﬁrm that our methods are more robust to outliers than the existing common classiﬁers.

@article{xu_robust_2018,
	title = {Robust {C}-loss kernel classifiers},
	volume = {29},
	issn = {2162-237X, 2162-2388},
	url = {https://ieeexplore.ieee.org/document/7801825/},
	doi = {10.1109/TNNLS.2016.2637351},
	abstract = {The correntropy-induced loss (C-loss) function has the nice property of being robust to outliers. In this paper, we study the C-loss kernel classiﬁer with the Tikhonov regularization term, which is used to avoid overﬁtting. After using the halfquadratic optimization algorithm, which converges much faster than the gradient optimization algorithm, we ﬁnd out that the resulting C-loss kernel classiﬁer is equivalent to an iterative weighted least square support vector machine (LS-SVM). This relationship helps explain the robustness of iterative weighted LS-SVM from the correntropy and density estimation perspectives. On the large-scale data sets which have low-rank Gram matrices, we suggest to use incomplete Cholesky decomposition to speed up the training process. Moreover, we use the representer theorem to improve the sparseness of the resulting C-loss kernel classiﬁer. Experimental results conﬁrm that our methods are more robust to outliers than the existing common classiﬁers.},
	language = {en},
	number = {3},
	urldate = {2023-12-01},
	journal = {IEEE Transactions on Neural Networks and Learning Systems},
	author = {Xu, Guibiao and Hu, Bao-Gang and Principe, Jose C.},
	month = mar,
	year = {2018},
	keywords = {/unread},
	pages = {510--522},
}

Downloads: 0

{"_id":"Yx7bKZiS7462jPDnm","bibbaseid":"xu-hu-principe-robustclosskernelclassifiers-2018","author_short":["Xu, G.","Hu, B.","Principe, J. C."],"bibdata":{"bibtype":"article","type":"article","title":"Robust C-loss kernel classifiers","volume":"29","issn":"2162-237X, 2162-2388","url":"https://ieeexplore.ieee.org/document/7801825/","doi":"10.1109/TNNLS.2016.2637351","abstract":"The correntropy-induced loss (C-loss) function has the nice property of being robust to outliers. In this paper, we study the C-loss kernel classiﬁer with the Tikhonov regularization term, which is used to avoid overﬁtting. After using the halfquadratic optimization algorithm, which converges much faster than the gradient optimization algorithm, we ﬁnd out that the resulting C-loss kernel classiﬁer is equivalent to an iterative weighted least square support vector machine (LS-SVM). This relationship helps explain the robustness of iterative weighted LS-SVM from the correntropy and density estimation perspectives. On the large-scale data sets which have low-rank Gram matrices, we suggest to use incomplete Cholesky decomposition to speed up the training process. Moreover, we use the representer theorem to improve the sparseness of the resulting C-loss kernel classiﬁer. Experimental results conﬁrm that our methods are more robust to outliers than the existing common classiﬁers.","language":"en","number":"3","urldate":"2023-12-01","journal":"IEEE Transactions on Neural Networks and Learning Systems","author":[{"propositions":[],"lastnames":["Xu"],"firstnames":["Guibiao"],"suffixes":[]},{"propositions":[],"lastnames":["Hu"],"firstnames":["Bao-Gang"],"suffixes":[]},{"propositions":[],"lastnames":["Principe"],"firstnames":["Jose","C."],"suffixes":[]}],"month":"March","year":"2018","keywords":"/unread","pages":"510–522","bibtex":"@article{xu_robust_2018,\n\ttitle = {Robust {C}-loss kernel classifiers},\n\tvolume = {29},\n\tissn = {2162-237X, 2162-2388},\n\turl = {https://ieeexplore.ieee.org/document/7801825/},\n\tdoi = {10.1109/TNNLS.2016.2637351},\n\tabstract = {The correntropy-induced loss (C-loss) function has the nice property of being robust to outliers. In this paper, we study the C-loss kernel classiﬁer with the Tikhonov regularization term, which is used to avoid overﬁtting. After using the halfquadratic optimization algorithm, which converges much faster than the gradient optimization algorithm, we ﬁnd out that the resulting C-loss kernel classiﬁer is equivalent to an iterative weighted least square support vector machine (LS-SVM). This relationship helps explain the robustness of iterative weighted LS-SVM from the correntropy and density estimation perspectives. On the large-scale data sets which have low-rank Gram matrices, we suggest to use incomplete Cholesky decomposition to speed up the training process. Moreover, we use the representer theorem to improve the sparseness of the resulting C-loss kernel classiﬁer. Experimental results conﬁrm that our methods are more robust to outliers than the existing common classiﬁers.},\n\tlanguage = {en},\n\tnumber = {3},\n\turldate = {2023-12-01},\n\tjournal = {IEEE Transactions on Neural Networks and Learning Systems},\n\tauthor = {Xu, Guibiao and Hu, Bao-Gang and Principe, Jose C.},\n\tmonth = mar,\n\tyear = {2018},\n\tkeywords = {/unread},\n\tpages = {510--522},\n}\n\n","author_short":["Xu, G.","Hu, B.","Principe, J. C."],"key":"xu_robust_2018","id":"xu_robust_2018","bibbaseid":"xu-hu-principe-robustclosskernelclassifiers-2018","role":"author","urls":{"Paper":"https://ieeexplore.ieee.org/document/7801825/"},"keyword":["/unread"],"metadata":{"authorlinks":{}},"html":""},"bibtype":"article","biburl":"https://bibbase.org/zotero/victorjhu","dataSources":["CmHEoydhafhbkXXt5"],"keywords":["/unread"],"search_terms":["robust","loss","kernel","classifiers","xu","hu","principe"],"title":"Robust C-loss kernel classifiers","year":2018}