Model Selection and the Principle of Minimum Description Length

By Mark H. Hansen, Bin Yu. 2001

845 citations recorded by the source corpus.

Subfields: Statistical and Nonlinear Physics, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/d6ac30380a592f96d1ec43e9cac1a560b3f4fbf4

How researchers see this paper

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

In Statistical and Nonlinear Physics

In Statistics and Probability

Citations in the corpus

FastSLAM: a factored solution to the simultaneous localization and mapping problem (2002, methodology): “and choose the value of Nt at every time step that corresponds with the Minimum Description Length (MDL) hypothesis [ 25 ].”
Prediction, learning, and games (2006, background): “For quite exhaustive surveys see Barron, Rissanen, Yu [22], Grünwald [134], and Hansen and Yu [143].”
Model Boosting for Spatial Weighting Matrix Selection in Spatial Lag Models (2010, background): “Bu« hlmann and Yu (2006) have shown that a data-driven compromise between AIC and BIC (Bayesian information criterion), namely the g-prior minimum description length (gMDL) introduced in Hansen and Yu (2001), can be successful in boosting for variable selection problems.”
A minimum description length approach to statistical shape modeling (2002, methodology): “Comparison of description lengths calculated using models from different classes can be used as a way of solving the model selection problem [17].”
Structural Break Estimation for Nonstationary Time Series Models (2006, neutral): “KEY WORDS: Changepoint; Genetic algorithm; Minimum description length principle; Nonstationarity.”
Model Selection (2018, methodology): “Apart from the expository articles already mentioned (Hocking (1976), Thompson (1978a,b), Amemiya (1980), Giles and Giles (1993), Hansen and Yu (2001), Rao and Wu (2001), Leeb and Pötscher (2005)), the article DeGooijer et al. (1985) provides a survey of model selection in time series analysis;…”
Catching up faster by switching sooner: a predictive approach to adaptive estimation with an application to the AIC–BIC dilemma (2012, methodology): “These include model meta-selection [De Luna and Skouras, 2003, Clarke, 1997], and Hansen and Yu’s gMDL version of MDL regression [Hansen and Yu, 2001]; also the “mongrel” procedure of Wong and Clarke [2004] has been designed to improve on Bayesian model averaging for small samples.”
A tutorial introduction to the minimum description length principle (2004, methodology): “We mention [Hansen and Yu 2000; Hansen and Yu 2001] reporting excellent behavior of MDL in regression contexts; and [Allen, Madani, and Greiner 2003; Kontkanen, Myllymäki, Silander, and Tirri 1999; Modha and Masry 1998] reporting excellent behavior of predictive (prequential) coding in Bayesian netw”
Mixtures of g-Priors in Generalized Linear Models (2015, methodology): “This includes the unit information prior where g = n (Kass and Wasserman 1995), and local and global empirical Bayes (EB) estimates of g (Copas 1983, 1997; Hansen and Yu 2001, 2003; Liang et al. 2008; Held, Sabanés Bové, and Gravestock 2015).”
Segmentation of Multivariate Mixed Data via Lossy Data Coding and Compression (2007, methodology): “Extensive simulation and experimental results of the proposed algorithm on synthetic and real data are given in Section V.”
Discussion: Performance of Bayesian Model Averaging (2003, methodology): “Hansen and Yu (2001) developed local empirical Bayes approaches in which the prior depends on the model in a data-dependentway, as in HC’s BLIC¤.”