Variable Selection

By Bastian Goldluecke, Variational Method, Velvety Reﬂectance, Video Mosaicing, Zhigang Zhu, Cees G. M. Snoek, Arnold M. W. Smeulders, Vasu Parameswaran, Ashok Veeraraghavan. 2019

Subfields: Statistics and Probability

How researchers see this paper

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

In Statistics and Probability

Citations in the corpus

On Security and Sparsity of Linear Classifiers for Adversarial Settings (2016, methodology): “We use here the elastic-net regularizer [25], combined with the hinge loss to obtain an SVM formulation with tunable sparsity: The level of sparsity can be tuned through the trade-oﬀ parameter λ ∈ (0 , 1).”
Learning Credible Models (2017, methodology): “…w ∗ | θ | [ i ] is the i th largest parameter sorted by magnitude; and ⊙ is the training time, Sun et al . explore using features identified as relevant Vapnik et al . explore the theory of learning with privileged information In the linear setting, the method simplifies to l 2 regularization on”
Automated versus Do-It-Yourself Methods for Causal Inference: Lessons Learned from a Data Analysis Competition (2017, methodology): “Variable selection techniques such as LASSO (Tibshirani, 1996) and the elastic net (Zou and Hastie, 2005) can help reduce the scope of the estimation problem to true confounders, so that more complicated algorithms can be used on the terms that really matter.10 Ensemble methods.”
DeepSynergy: predicting anti-cancer drug synergy with Deep Learning (2017, methodology): “Supplementary Table S3 summarizes the considered hyper-parameter ranges for Elastic Nets.”
mgm: Estimating Time-Varying Mixed Graphical Models in High-Dimensional Data (2015, background): “Via the argument alphaSeq one can specify any convex combination of the `1 and `2 penality (the elastic net penality, see Zou and Hastie (2005)). alphaSeq = 1 corresponds to the `1-penalty (default) and alphaSeq = 0 to the `2 penality.”
Thirty Years of Machine Learning: The Road to Pareto-Optimal Wireless Networks (2019, methodology): “Moreover, ridge regression [192], least absolute shrinkage and selection operator (LASSO) regression [193] and ElasticNet regression [194] are widely applied, when independent variables are of multi-collinear nature and highly correlated.”
Distributed Optimization for Non-Strongly Convex Regularizers (2016, background): “17 2.4 Convergence Analysis for Smooth Losses . . . . . . . . . . . . . 18 2.5 Elastic Net . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20”
Feature selection in machine learning: A new perspective (2018, methodology): “Lasso-based feature selection methods have well stability, including Lasso for regression model [57], Lars [55], Adaptive-lasso [58], elastic net [59], and so on.”
Learning Models with Uniform Performance via Distributionally Robust Optimization (2018, methodology): “We use the absolute loss `(θ; (x, y)) = |θ>x− y| and compare method (2) with constrained forms of lasso, ridge, and elastic net regularization [102], taking constraints of the form”
Uncertainty and crude oil market volatility: new evidence (2020, methodology): “Many studies have documented that the elastic net and lasso generally improve prediction accuracy because of bias-variance tradeoff (see, e.g. Zou and Hastie 2005; Li and Tsiakas 2017).”
Sifting and winnowing: Analysis of farmer field data for soybean in the US North-Central region (2018, methodology): “For example, regression procedures such as LASSO, LAR, and elastic net have properties that can mitigate data multi-collinearity issues (Zou and Hastie, 2005, Dormann et al., 2013).”