Quadratic Convergence of Smoothing Newton's Method for 0/1 Loss Optimization

DOI10.1137/21M1409445zbMath1483.90126arXiv2103.14987MaRDI QIDQ5020852

Sheng-Long Zhou, Li-Li Pan, Hou-Duo Qi, Nai-Hua Xiu

Publication date: 7 January 2022

Published in: SIAM Journal on Optimization (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/2103.14987

zbMATH Keywords

Newton's methodoptimality conditionslocally quadratic convergence\(0/1\) loss functionsuperior numerical performance

Mathematics Subject Classification ID

Numerical mathematical programming methods (65K05) Nonconvex programming, global optimization (90C26) Nonlinear programming (90C30)

Related Items

Curved elements in weak Galerkin finite element methods ⋮ A majorization penalty method for SVM with sparse constraint

Uses Software

• LIBSVM
• LIBLINEAR
• ElemStatLearn

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Supersparse linear integer models for optimized medical scoring systems
• A smoothing Newton-type algorithm of stronger convergence for the quadratically constrained convex quadratic programming
• Noisy 1-bit compressive sensing: models and algorithms
• Analysis of the consistency of a mixed integer programming-based multi-category constrained discriminant model
• Non-parametric analysis of a generalized regression model. The maximum rank correlation estimator
• Solving mixed integer classification problems by decomposition
• On the difficulty of approximately maximizing agreements.
• A smoothing Newton method for general nonlinear complementarity problems
• Support-vector networks
• Regularization networks and support vector machines
• Semi-smooth Newton methods for state-constrained optimal control problems
• On the Minimization Over Sparse Symmetric Sets: Projections, Optimality Conditions, and Algorithms
• Improved Iteratively Reweighted Least Squares for Unconstrained Smoothed $\ell_q$ Minimization
• Sparsity Constrained Nonlinear Optimization: Optimality Conditions and Algorithms
• Binarized Support Vector Machines
• Robust 1-Bit Compressive Sensing via Binary Stable Embeddings of Sparse Vectors
• Support Vector Machines with the Ramp Loss and the Hard Margin Loss
• Robust Decoding from 1-Bit Compressive Sampling with Ordinary and Regularized Least Squares
• Robust Truncated Hinge Loss Support Vector Machines
• Integer Programming Solution of a Classification Problem
• Global and superlinear convergence of the smoothing Newton method and its application to general box constrained variational inequalities
• The Primal-Dual Active Set Strategy as a Semismooth Newton Method
• Robust 1-bit Compressive Sensing Using Adaptive Outlier Pursuit
• A Primal Dual Active Set Algorithm With Continuation for Compressed Sensing
• Agnostic Learning of Monomials by Halfspaces Is Hard

This page was built for publication: Quadratic Convergence of Smoothing Newton's Method for 0/1 Loss Optimization