Large-scale logistic regression and linear support vector machines using spark

By Chieh-Yen Lin, Cheng-Hao Tsai, Ching-Pei Lee, Chih-Jen Lin. 2014

86 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Federated Optimization: Distributed Machine Learning for On-Device Intelligence (2016, methodology): “The DiSCO algorithm is related to [52, 106], a distributed truncated Newton method.”
Communication-Efficient Distributed Optimization of Self-Concordant Empirical Loss (2015, background): “Under the assumption that each local function fi is L-smooth and λ-strongly convex, the ADMM approach can achieve linear convergence, and the best known complexity is O(√κ log(1/ǫ)) [17].”
Stochastic, Distributed and Federated Optimization for Machine Learning (2017, methodology): “The DiSCO algorithm is related to [98, 202], a distributed truncated Newton method.”
DiSCO: Distributed Optimization for Self-Concordant Empirical Loss (2015, methodology): “A similar approach was proposed by Zhuang et al. (2014) and Lin et al. (2014) for ERM of linear predictors, where they use a distributed truncated Newton method and conjugate gradient (CG) method for solving the Newton system.”
Recent Advances of Large-Scale Linear Classification (2012, methodology): “Newton: the Newton method discussed in Section IV-C has been parallelized in, for example, [134], [135].”
Distributed Box-Constrained Quadratic Optimization for Dual Linear SVM (2015, unknown): “Because the main computations in batch solvers for (1) are matrix-vector products that can be naively parallelized, several works have successfully adapted these solvers to distributed environments (Agarwal et al., 2014; Zhang et al., 2012; Zhuang et al., 2015; Lin et al., 2014).”
Distributed Newton Methods for Regularized Logistic Regression (2015, methodology): “[12] implement a variant of our approach on Spark1 through private communication, but they mainly focus on the efficient use of Spark.”
A Distributed Quasi-Newton Algorithm for Empirical Risk Minimization with Nonsmooth Regularization (2018, methodology): “Other modified batch methods that utilize the Hessian of the objective in various ways are also communication-efficient under their own additional assumptions [7, 12, 20, 26, 28].”