3,613 Hits in 4.6 sec

Nonconvex Regularizations for Feature Selection in Ranking With Sparse SVM

Lea Laporte, Remi Flamary, Stephane Canu, Sebastien Dejean, Josiane Mothe
2014 IEEE Transactions on Neural Networks and Learning Systems  
In this work, we propose a general framework for feature selection in learning to rank using SVM with a sparse regularization term.  ...  Feature selection in learning to rank has recently emerged as a crucial issue.  ...  Sparse regularized SVM for preferences ranking To achieve feature selection in the context of SVM, a common solution is to introduce a sparse regularization term.  ... 
doi:10.1109/tnnls.2013.2286696 fatcat:li45e4zbxbbp3hgkz2uzdcngpq

Direct convex relaxations of sparse SVM

Antoni B. Chan, Nuno Vasconcelos, Gert R. G. Lanckriet
2007 Proceedings of the 24th international conference on Machine learning - ICML '07  
We propose two direct, novel convex relaxations of a nonconvex sparse SVM formulation that explicitly constrains the cardinality of the vector of feature weights.  ...  all available features in the input space.  ...  Acknowledgments The authors thank the anonymous reviewers for insightful comments, and Sameer Agarwal for helpful discussions.  ... 
doi:10.1145/1273496.1273515 dblp:conf/icml/ChanVL07 fatcat:apgkb6fgwzgo5lri3fdaunsb74

Sparse Support Vector Infinite Push [article]

Alain Rakotomamonjy
2012 arXiv   pre-print
In this paper, we address the problem of embedded feature selection for ranking on top of the list problems.  ...  We pose this problem as a regularized empirical risk minimization with p-norm push loss function (p=∞) and sparsity inducing regularizers.  ...  so as to perform feature selection in a top-ranking learning problem.  ... 
arXiv:1206.6432v1 fatcat:kfkq4ib2gzevvcuyt5z7nvnwda

Learning to rank using 1-norm regularization and convex hull reduction

Xiaofei Nan, Yixin Chen, Xin Dang, Dawn Wilkins
2010 Proceedings of the 48th Annual Southeast Regional Conference on - ACM SE '10  
We also propose a 1-norm regularization approach to simultaneously find a linear ranking function and to perform feature subset selection. The proposed method is formulated as a linear program.  ...  We present in this paper a convex hull reduction method to reduce this impact.  ...  ACKNOWLEDGMENTS Xiaofei Nan, Yixin Chen, and Dawn Wilkins were supported in part by the US National Science Foundation under award number EPS-0903787.  ... 
doi:10.1145/1900008.1900052 dblp:conf/ACMse/NanCDW10 fatcat:tzkeynse5jg2bkebyc7osgycdq

Primal explicit max margin feature selection for nonlinear support vector machines

Aditya Tayal, Thomas F. Coleman, Yuying Li
2014 Pattern Recognition  
Embedding feature selection in nonlinear SVMs leads to a challenging non-convex minimization problem, which can be prone to suboptimal solutions.  ...  We devise an alternating optimization approach to tackle the problem efficiently, breaking it down into a convex subproblem, corresponding to standard SVM optimization, and a non-convex subproblem for  ...  ∞ < tol We can use any convex solver for the SVM subproblem and use the bound-constrained trust-region algorithm described in Section 3.1 to solve the non-convex feature selection subproblem.  ... 
doi:10.1016/j.patcog.2014.01.003 fatcat:tf6qe2abpjewbkectfyoogvoci

Multiple Indefinite Kernel Learning for Feature Selection

Hui Xue, Yu Song, Hai-Ming Xu
2017 Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence  
In the algorithm, we reformulate the non-convex optimization problem of primal IKSVM as a difference of convex functions (DC) programming and transform the non-convex problem into a convex one with the  ...  Multiple kernel learning for feature selection (MKL-FS) utilizes kernels to explore complex properties of features and performs better in embedded methods.  ...  Tan et al. focused on sparse support vector machines (SVM) with l 0 -norm whose convex relaxation can be further formulated as an MKL problem, where each kernel corresponds to a sparse feature subset  ... 
doi:10.24963/ijcai.2017/448 dblp:conf/ijcai/XueSX17 fatcat:z2mgcyayiff3nbreltgsjpxzjm

Learning Sparse SVM for Feature Selection on Very High Dimensional Datasets

Mingkui Tan, Li Wang, Ivor W. Tsang
2010 International Conference on Machine Learning  
A sparse representation of Support Vector Machines (SVMs) with respect to input features is desirable for many applications.  ...  In this paper, by introducing a 0-1 control variable to each input feature, l 0 -norm Sparse SVM (SSVM) is converted to a mixed integer programming (MIP) problem.  ...  Acknowledgments This research was in part supported by Singapore MOE AcRF Tier-1 Research Grant (RG15/08).  ... 
dblp:conf/icml/TanWT10 fatcat:qsmntopnm5hehkkm4vdl7j6pcy

Self-calibrated Brain Network Estimation and Joint Non-Convex Multi-Task Learning for Identification of Early Alzheimer's Disease

Baiying Lei, Nina Cheng, Alejandro F Frangi, Ee-Leng Tan, Jiuwen Cao, Peng Yang, Ahmed Elazab, Jie Du, Yanwu Xu, Tianfu Wang
2020 Medical Image Analysis  
The learning process is completed by non-convex regularizer, which effectively reduces the penalty bias of trace norm and approximates the original rank minimization problem.  ...  Finally, the most relevant disease features classified using a support vector machine (SVM) for MCI identification.  ...  In the feature selection stage, non-convex regularizer is used to complete the subspace learning of samples.  ... 
doi:10.1016/ pmid:32059169 fatcat:w6qno5ofufdpjdipqjardolkhe

Regularization and feature selection for large dimensional data [article]

Nand Sharma, Prathamesh Verlekar, Rehab Ashary, Sui Zhiquan
2019 arXiv   pre-print
The focus of our research here are five embedded feature selection methods which use either the ridge regression, or Lasso regression, or a combination of the two in the regularization part of the optimization  ...  Feature selection has evolved to be an important step in several machine learning paradigms.  ...  The top ranked 1000 features are selected for model selection of SVM using 5-fold crossvalidation on the datasets comprising both training and validation sets. 4.  ... 
arXiv:1712.01975v3 fatcat:leq5e6mzb5c37omybgexwiel6a

Sparse Logistic Regression with Lp Penalty for Biomarker Identification

Zhenqiu Liu, Feng Jiang, Guoliang Tian, Suna Wang, Fumiaki Sato, Stephen J. Meltzer, Ming Tan
2007 Statistical Applications in Genetics and Molecular Biology  
In this paper, we propose a novel method for sparse logistic regression with non-convex regularization Lp (p <1).  ...  Biomarkers identified with our methods are compared with that in the literature. Our computational results show that Lp Logistic regression (p <1) outperforms the L1 logistic regression and SCAD SVM.  ...  Bradley and Mangasarian (1998) proposed the L 1 SVM for feature selection.  ... 
doi:10.2202/1544-6115.1248 pmid:17402921 fatcat:sbdavwwabravxjdmndfn643jr4

Similarity Learning for High-Dimensional Sparse Data [article]

Kuan Liu and Aurélien Bellet and Fei Sha
2019 arXiv   pre-print
The core idea is to parameterize the similarity measure as a convex combination of rank-one matrices with specific sparsity structures.  ...  In this paper, we propose a method that can learn efficiently similarity measure from high-dimensional sparse data.  ...  Acknowledgments This work was in part supported by the Intelligence Advanced Research Projects Activity (IARPA) via Department of Defense U.S.  ... 
arXiv:1411.2374v3 fatcat:qjlg5iyz6vbbbjgqdeyaifsoci

Sturm: Sparse Tubal-Regularized Multilinear Regression for fMRI [article]

Wenwen Li, Jian Lou, Shuo Zhou, Haiping Lu
2018 arXiv   pre-print
In this work, we study t-SVD for sparse multilinear regression and propose a Sparse tubal-regularized multilinear regression (Sturm) method for fMRI.  ...  Recent sparse multilinear regression methods based on tensor are emerging as promising solutions for fMRI, yet existing works rely on unfolding/folding operations and a tensor rank relaxation with limited  ...  In Lasso + SVM, ENet + SVM, Remurs + SVM, and Sturm + SVM, we rank the selected features by their associated absolute values of W in the descending order and feed the top η% of the features to SVM.  ... 
arXiv:1812.01496v1 fatcat:llyqce27jrdgba3j7dkk6nofgm

Feature selection in machine learning: an exact penalty approach using a Difference of Convex function Algorithm

Hoai An Le Thi, Hoai Minh Le, Tao Pham Dinh
2014 Machine Learning  
We develop an exact penalty approach for feature selection in machine learning via the zero-norm 0 -regularization problem.  ...  The algorithm is implemented for feature selection in SVM, that requires solving one linear program at each iteration and enjoys interesting convergence properties.  ...  , feature selection in learning to rank with sparse SVM, etc.  ... 
doi:10.1007/s10994-014-5455-y fatcat:b4bnlo4rcbhwhgjnuhpa5wcfce

$\ell_{p}-\ell_{q}$ Penalty for Sparse Linear and Sparse Multiple Kernel Multitask Learning

A. Rakotomamonjy, R. Flamary, G. Gasso, S. Canu
2011 IEEE Transactions on Neural Networks  
Then, for the more general case when 0 < p < 1, we solve the resulting non-convex problem through a majorization-minimization approach.  ...  Our contribution in this context is to provide an efficient scheme for computing the ℓ1 − ℓq proximal operator.  ...  ALGORITHMS FOR JOINTLY SPARSE MULTI-TASK SVM In this section, we propose some algorithms for solving the sparse multi-task SVM problem when using Ω p,q as a regularizer with values p ≤ 1 and 1 ≤ q ≤ 2.  ... 
doi:10.1109/tnn.2011.2157521 pmid:21813358 fatcat:cdjradqjhrgdjmozxo4crrlhra

Multi-Task Joint Sparse and Low-Rank Representation for the Scene Classification of High-Resolution Remote Sensing Image

Kunlun Qi, Wenxuan Liu, Chao Yang, Qingfeng Guan, Huayi Wu
2016 Remote Sensing  
In this paper, we introduce a multi-task joint sparse and low-rank representation model to combine the strength of multiple features for HRS image interpretation.  ...  The proposed model is optimized as a non-smooth convex optimization problem using an accelerated proximal gradient method.  ...  The problem is intractable for the two non-smooth convex regularization terms P(W) and Q(W).  ... 
doi:10.3390/rs9010010 fatcat:phdn4lsq5zhtbjonmsh3taf73q
« Previous Showing results 1 — 15 out of 3,613 results