Trust Region Newton Method for Logistic Regression

By Chih-Jen Lin, Ruby Chiu-Hsing Weng, S. Sathiya Keerthi. 2008

347 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Optimization & Training

Citations in the corpus

Fast large-scale optimization by unifying stochastic gradient and quasi-Newton methods (2013, methodology): “In (Lin et al., 2008) a trust region Newton method is used to train logistic regression and linear SVMs using minibatches.”
Hyperparameter optimization with approximate gradient (2016, methodology): “Efﬁcient schemes for multiplication by the Hessian can be derived for least squares, logistic regression (Lin et al., 2008) and other general loss functions (Pearlmutter, 1994).”
TILDE: A Temporally Invariant Learned DEtector (2014, methodology): “To estimate the hyperplane to add, we apply a trust region Newton method [22], as in the widely-used LibLinear library [9].”
Preconditioned Stochastic Gradient Langevin Dynamics for Deep Neural Networks (2015, methodology): “We then test BLR on a large-scale Adult dataset, a9a (Lin, Weng, and Keerthi 2008), with N train = 32561 , N test = 16281 , and D = 123 .”
Distributed Newton Methods for Regularized Logistic Regression (2015, methodology): “In [Zhang, Lee, and Shin, 2012], they consider a coordinate descent method [Hsieh et al., 2008] and a trust region Newton method [Lin, Weng, and Keerthi, 2008] from the software LIBLINEAR [Fan et al., 2008].”
A General Iterative Shrinkage and Thresholding Algorithm for Non-convex Regularized Optimization Problems (2013, background): “Two of them (news20, realsim) have been preprocessed as two-class data sets (Lin et al., 2008).”
Towards Better Representations with Deep/Bayesian Learning (2018, methodology): “We then test BLR on a large-scale Adult dataset, a9a (Lin et al., 2008), with Ntrain “ 32561, Ntest “ 16281, and D “ 123.”
A Comparison of Optimization Methods and Software for Large-scale L1-regularized Linear Classification (2010, methodology): “Lin et al. (2008) applied TRON to solve L2-regularized logistic regression and showed that TRON outperforms LBFGS for document data.”
Dual coordinate descent methods for logistic regression and maximum entropy models (2011, methodology): “TRON: a trust region Newton method for logistic regression (Lin et al., 2008).”
Bundle Methods for Regularized Risk Minimization (2010, methodology): “In the case of logistic regression, we compareBMRM to the state of the art trust region Newton method for logistic regression (Lin et al., 2008) which is also available in theLIBLINEAR package (option “-s 0”).”
Sub-Sampled Newton Methods II: Local Convergence Rates (2016, positive): “Such second order information have long been used in many machine learning applications [6, 51, 26, 30, 11, 12] as well as many PDE constrained inverse problems [15, 41, 40, 10, 19].”
Sub-Sampled Newton Methods I: Globally Convergent Algorithms (2016, background): “Such second order information have long been used in many machine learning applications [6, 51, 27, 31, 10, 11].”
Newton-Stein Method: An Optimization Method for GLMs via Stein's Lemma (2015, background): “Set t = 0 and sub-sample a set of indices S ⊂ [ n ] uniformly at random.”
Large-scale Multi-label Learning with Missing Labels (2013, methodology): “1Note that although L2-hinge loss is not twice-differentiable, the sub-differential of ∂∂b `(a, b) still can be used for TRON to solve (7).”