Adaptive Model Selection

By Xiaotong Shen, Jianming Ye. 2002

219 citations recorded by the source corpus.

Subfields: Management Science and Operations Research, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/0d904bc6b5a9c27bc4fe5f64b725dbdb24075863

How researchers see this paper

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

In Management Science and Operations Research

In Statistics and Probability

Citations in the corpus

Optimal Model Assessment, Selection, and Combination (2006, methodology): “In some situations, such as in wavelet thresholding (Shen and Ye 2002), it is possible to derive an analytic expression.”
Efficient Empirical Bayes Variable Selection and Estimation in Linear Models (2005, methodology): “A number of other variable selection methods have been introduced in recent years (George and McCulloch 1993; Foster and George 1994; Breiman 1995; Tibshirani 1996; Fan and Li 2001; Shen and Ye 2002; Efron, Johnston, Hastie, and Tibshirani 2004).”
Stein Unbiased GrAdient estimator of the Risk (SUGAR) for Multiple Parameter Selection (2014, unknown): “5 Finite-Difference SURE An alternative initiated in [60, 71] and rediscovered in [51] consists in estimating tr (A∂1μ(y, θ)A ∗) via finite differences given, for ε > 0, by”
Optimal Geostatistical Model Selection (2007, methodology): “5, as was done by Shen and Ye (2002) and Shen, Huang, and Ye (2004) by trading off some bias for lower variance. These authors showed in some linear and generalized linear regression examples that model selection using the data perturbation approach is not sensitive to the choice of τ . Shen and Hua”
Towards data driven selection of a penalty function for data driven Neyman tests (2006, methodology): “Since small penalties are associated with detecting “large” models, again using the terminology of Shen and Ye [25], the effect is that the related tests are less powerful for small departures (from the null hypothesis) than those related to BIC.”
Efficient Estimation of Semiparametric Multivariate Copula Models (2006, unknown): “In principle, we could use any model selection methodology, including cross-validation (see, e.g., Coppejans and Gallant 2002), covariance penalty (see, e.g., Shen and Ye 2002), and others, to choose the number of terms, K nj , in the sieve approximation.”
Penalized Model-Based Clustering with Application to Variable Selection (2007, methodology): “Although other resampling-based model selection methods, such as cross-validation or generalized degrees of freedom (Shen and Ye, 2002) can be employed, they are computationally more demanding, and even prohibitive for large and/or high-dimensional data as considered here.”
On the non‐negative garrotte estimator (2007, methodology): “Various other variable selection methods have been introduced in recent years (George and McCulloch, 1993; Foster and George, 1994; Breiman, 1995; Tibshirani, 1996; George and Foster, 2000; Fan and Li, 2001; Shen and Ye, 2002; Efron et al., 2004; Yuan and Lin, 2005; Zou and Hastie, 2005).”
Model selection and estimation in regression with grouped variables (2006, methodology): “…is Cp.µ̂/= ‖Y − µ̂‖ 2 σ2 −n+2 dfµ,σ2 , .6:1/ where dfµ,σ2 = n∑ i=1 cov.µ̂i, Yi/=σ2: .6:2/ Since the definition of the degrees of freedom involves the unknowns, in practice, it is often estimated through the bootstrap (Efron et al., 2004) or some data perturbation methods (Shen and Ye, 2002).”
The Adaptive Lasso and Its Oracle Properties (2006, background): “Furthermore, these selection procedures ignore the stochastic errors or uncertainty in the variable selection stage (Fan and Li 2001; Shen and Ye 2002).”
Minimal Penalties for Gaussian Model Selection (2007, background): “Another approach is developed in Shen and Ye [36] who consider penalties of the form λε2D and estimate λ by the minimization of some estimated risk.”
On the “degrees of freedom” of the lasso (2007, methodology): “To overcome the analytical difficulty, Ye [24] and Shen and Ye [20] proposed using a data-perturbation technique to numerically compute an (approximately) unbiased estimate for df (μ̂) when the analytical form of μ̂ is unavailable.”
Cross-validation for selecting a model selection procedure (2015, methodology): “Shen and Ye (2002) and Zhang (2009) propose adaptive model selection methods by introducing data-driven penalty coefficients into information criteria.”
Parametric or nonparametric? A parametricness index for model selection (2011, background): “Eﬀorts have been put on using penalties that are data-dependent and adaptive (see, e.g., [7, 31, 34, 38, 55, 56, 68]).”