An inexact successive quadratic approximation method for L-1 regularized optimization.
Richard H. ByrdJorge NocedalFigen ÖztoprakPublished in: Math. Program. (2016)
Keyphrases
- objective function
- optimization method
- pairwise
- cost function
- high accuracy
- optimization algorithm
- approximation methods
- relative error
- synthetic data
- constrained optimization
- em algorithm
- feature extraction
- significant improvement
- high precision
- classification method
- risk minimization
- optimization model
- data sets
- cross validation
- segmentation method
- support vector machine svm
- theoretical analysis
- simulated annealing
- denoising
- support vector
- bayesian networks
- similarity measure