An efficient augmented Lagrangian method for support vector machine
From MaRDI portal
Publication:5135259
Abstract: Support vector machine (SVM) has proved to be a successful approach for machine learning. Two typical SVM models are the L1-loss model for support vector classification (SVC) and -L1-loss model for support vector regression (SVR). Due to the nonsmoothness of the L1-loss function in the two models, most of the traditional approaches focus on solving the dual problem. In this paper, we propose an augmented Lagrangian method for the L1-loss model, which is designed to solve the primal problem. By tackling the nonsmooth term in the model with Moreau-Yosida regularization and the proximal operator, the subproblem in augmented Lagrangian method reduces to a nonsmooth linear system, which can be solved via the quadratically convergent semismooth Newton's method. Moreover, the high computational cost in semismooth Newton's method can be significantly reduced by exploring the sparse structure in the generalized Jacobian. Numerical results on various datasets in LIBLINEAR show that the proposed method is competitive with the most popular solvers in both speed and accuracy.
Recommendations
Cites work
- A highly efficient semismooth Newton augmented Lagrangian method for solving lasso problems
- A Newton-CG augmented Lagrangian method for semidefinite programming
- A nonsmooth version of Newton's method
- A Quadratically Convergent Newton Method for Computing the Nearest Correlation Matrix
- A semismooth Newton method for support vector classification and regression
- A semismooth Newton method for the nearest Euclidean distance matrix problem
- A sequential semismooth Newton method for the nearest low-rank correlation matrix problem
- A unified formulation and fast accelerated proximal gradient method for classification
- An efficient inexact ABCD method for least squares semidefinite programming
- An implementable proximal point algorithmic framework for nuclear norm minimization
- Augmented Lagrangians and Applications of the Proximal Point Algorithm in Convex Programming
- Computing the nearest Euclidean distance matrix with low embedding dimensions
- Exponentiated gradient algorithms for conditional random fields and max-margin Markov networks
- scientific article; zbMATH DE number 50640 (Why is no real title available?)
- scientific article; zbMATH DE number 3204910 (Why is no real title available?)
- Inexact semismooth Newton methods for large-scale complementarity problems
- Large-scale linear support vector regression
- LIBLINEAR: a library for large linear classification
- Methods of conjugate gradients for solving linear systems
- Monotone Operators and the Proximal Point Algorithm
- Online Learning with Kernels
- Pegasos: primal estimated sub-gradient solver for SVM
- Proximité et dualité dans un espace hilbertien
- Regularized nonsmooth Newton method for multi-class support vector machines
- SDPNAL+: a majorized semismooth Newton-CG augmented Lagrangian method for semidefinite programming with nonnegative constraints
- SDPNAL+: A Matlab software for semidefinite programming with bound constraints (version 1.0)
- Semismooth and Semiconvex Functions in Constrained Optimization
- Semismooth Matrix-Valued Functions
- Solving the OSCAR and SLOPE models using a semismooth Newton-based augmented Lagrangian method
- Support-vector networks
- Training a Support Vector Machine in the Primal
- Two smooth support vector machines for -insensitive regression
Cited in
(8)- Smooth augmented Lagrangian method for twin bounded support vector machine
- Bilevel hyperparameter optimization for support vector classification: theoretical analysis and a solution method
- Proximal gradient method for huberized support vector machine
- A semismooth Newton method for support vector classification and regression
- Analysis of loss functions in support vector machines
- scientific article; zbMATH DE number 7112777 (Why is no real title available?)
- A majorization penalty method for SVM with sparse constraint
- An L₂ regularization reduced quadratic surface support vector machine model
This page was built for publication: An efficient augmented Lagrangian method for support vector machine
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5135259)
