A geometric approach to leveraging weak learners

Theoretical Computer Science

Volumn 284, Issue 1, 2002, Pages 67-108

  Duffy, Nigel ^a   Helmbold, David ^a  

^a UNIVERSITY OF CALIFORNIA   (United States)

Author keywords

Boosting; Classification; Ensemble methods; Gradient descent; Learning

Indexed keywords

COMPUTATIONAL GEOMETRY; COMPUTER SCIENCE; GRADIENT METHODS; LEARNING ALGORITHMS;

ENSEMBLE METHODS;

LEARNING SYSTEMS;

EID: 0037031029     PISSN: 03043975     EISSN: None     Source Type: Journal    
DOI: 10.1016/S0304-3975(01)00083-4     Document Type: Conference Paper

References (38)

1. Polynomial learnability of probabilistic concepts with respect to the Kullback-Leibler divergence
2. An empirical comparison of voting classification algorithms: Bagging, boosting and variants
3. A training algorithm for optimal margin classifiers
4. Bagging predictors
5. Prediction games and arcing algorithms
6. Supervised and unsupervised discretization of continuous features
7. Potential boosters?
8. Boosting a weak learning algorithm by majority
9. An adaptive version of the boost-by-majority algorithm
10. Experiments with a new boosting algorithm
11. A decision-theoretic generalization of online learning and an application to boosting
12. General convergence results for linear discriminant updates
13. Equivalence of models for polynomial learnability
14. Cryptographic limitations on learning Boolean formulae and finite automata
15. Additive versus exponentiated gradient updates for linear prediction
16. Boosting as entropy projection
17. Additive models, boosting, inference for generalized divergences
18. Direct optimization of margins improves generalization in combined classifiers
19. Boosting algorithms as gradient descent
20. Bagging, boosting and C4.5
21. Drifting games
22. Boosting the margin: A new explanation for the effectiveness of voting methods
23. Improved boosting algorithms using confidence-rated predictions
24. Margin distribution bounds on generalization
25. A theory of the learnable