Metric-based methods for adaptive model selection and regularization

Machine Learning

Volumn 48, Issue 1-3, 2002, Pages 51-84

  Schuurmans, Dale ^a   Southey, Finnegan ^a  

^a UNIVERSITY OF WATERLOO   (Canada)

Author keywords

Model selection; Regularization; Unlabeled examples

Indexed keywords

ADAPTIVE CONTROL SYSTEMS; COMPUTATIONAL COMPLEXITY; ERROR ANALYSIS; MATHEMATICAL MODELS; MAXIMUM LIKELIHOOD ESTIMATION; PROBABILITY DENSITY FUNCTION; PROBABILITY DISTRIBUTIONS; REGRESSION ANALYSIS;

CONDITIONAL DENSITY ESTIMATION; ERROR ESTIMATION; METRIC BASED METHODS; MODEL SELECTION; REGULARIZATION; SUPERVISED LEARNING; UNLABELLED EXAMPLES;

LEARNING SYSTEMS;

EID: 0036643079     PISSN: 08856125     EISSN: None     Source Type: Journal    
DOI: 10.1023/A:1013947519741     Document Type: Article

References (40)

1. Statistical predictor information
2. A new look at the statistical model identification
3. Combining labeled and unlabeled data with co-training
4. Bagging predictors
5. The relative value of labeled and unlabeled samples in pattern recognition with an unknown mixing parameter
6. Comparison of VC-method with classical methods for model selection
7. Smoothing noisy data with spline functions
8. Computers and the theory of statistics: Thinking the unthinkable
9. The risk inflation criterion for multiple regression
10. A decision-theoretic generalization of on-line learning and an application to boosting
11. Applications of model selection techniques to polynomial approximation
12. Robust classifiers by mixed adaptation
13. A comparison of scientific and engineering criteria for Bayesian model selection
14. A study of cross-validation and bootstrap for accuracy estimation and model selection
15. Neural network ensembles, cross validation, and active learning
16. A data-dependent skeleton estimate for learning
17. Bayesian interpolation
18. p
19. A mixture of experts classifier with learning based on both labelled and unlabelled data
20. Normal discrimination with unclassified observations
21. Generating accurate and diverse members of a neural-network ensemble
22. Regularization algorithms for learning that are equivalent to multilayer networks
23. Learning from a mixture of labeled and unlabeled examples with parametric side information
24. Stochastic complexity and modeling
25. Overfitting avoidance as bias
26. A conservation law for generalization performance
27. A new metric-based approach to model selection
28. An adaptive regularization criterion for supervised learning
29. Characterizing the generalization performance of model selection strategies
30. Estimating the dimension of a model
31. An optimal selection of regression variables
32. Using unlabeled data for supervised learning
33. A Bayesian approach to prediction using polynomials