Model selection for probabilistic clustering using cross-validated likelihood

By Padhraic Smyth. Statistics and computing, 2000

352 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Epidemiology, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/d2d309fad74402582786fd05214f8b988efd132e

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 Epidemiology

In Statistics and Probability

Citations in the corpus

Survey of clustering algorithms (2005, background): “[114], [216], cross validation-based information criterion (CVIC) [254] and covariance inflation criterion (CIC) [272], with their characteristics, are summarized in [197].”
Model-Based Clustering, Discriminant Analysis, and Density Estimation (2002, methodology): “…approaches include an informational complexity criterion called ICOMP (Bozdogan 1994), an entropy criterion called NEC (Celeux and Soromenho 1996; Biernacki, Celeux, and Govaert 1999), the integrated classi cation likelihood (Biernacki et al. 2000), and cross-validated likelihood (Smyth 2000).”
Probabilistic clustering of extratropical cyclones using regression mixture models (2007, methodology): “This calculation was repeated C = 10 times over multiple training-test partitions of the data to generate average out-of-sample log-probability (or loglikelihood ) scores (Smyth 2000; Smyth et al. 1999).”
Clustering aggregation (2005, methodology): “The second approach uses model selection methods, for example, Bayesian information criterion (BIC) [Schwarz 1978], or cross-validated likelihood [Smyth 2000] to compare models with different numbers of clusters.”
Unified Cross-Validation Methodology For Selection Among Estimators and a General Cross-Validated Adaptive Epsilon-Net Estimator: Finite Sample Oracle Inequalities and Examples (2003, neutral): “Recent work on (V -fold or Monte-Carlo) cross-validated likelihood methods for choosing the number of components in mixture models is found in Smyth (2000) and Pavlic and van der Laan (2003).”
Assessing a Mixture Model for Clustering with the Integrated Completed Likelihood (2000, methodology): “(Numerical experiments comparing BIC and CS criteria are in [3] and numerical experiments comparing BIC, MML, and their Bayesian criterion are in [25].) An alternative promising approach for estimating the proper number of clusters is the cross-validated likelihood approach suggested by Smyth [ 28 ]”
Sourcebook of parallel computing (2003, methodology): “The principal component and hidden Markov classification algorithms currently used in the earthquake field [481, 786, 816, 817, 877, 932, 944] illustrate the diversity in data-mining methods when compared with the decision-tree methods of Section 8.”
Strictly Proper Scoring Rules, Prediction, and Estimation (2007, methodology): “Smyth (2000) used the log-likelihood as the criterion function in cross-validation, as here, calling the resulting method cross-validated likelihood, but used a fixed holdout sample size.”
A Survey of Clustering Data Mining Techniques (2006, methodology): “Smyth [203] suggested a likelihood cross-validation technique for determination the best k.”
Quantifying Uncertainty in Batch Personalized Sequential Decision Making (2014, methodology): “In more detail, for each class, for each state–action pair (s, a), we define Dir(λ(s,a)), where λ(s,a) is a count vector for each next state s′.”
Merging Seasonal Rainfall Forecasts from Multiple Statistical Models through Bayesian Model Averaging (2012, background): “(3) with a cross-validation likelihood function (Rust and Schmittlein 1985; Shinozaki et al. 2010; Smyth 1996, 2000; Stone 1977).”
Testing for Zero Inflation in Count Models: Bias Correction for the Vuong Test (2013, neutral): “The KLD is a measure of the distance between two probability distributions, which is the basis of many measures used for model comparison and selection, including the AIC (Akaike 1974), the Takeuchi information criterion (Konishi and Kitagawa 1996), the generalized information criterion (Konishi and”