Shapire, R.E.: The Strength of Weak Learnability. Machine Learning 5 (1990) 197--227 Home/Search Document Not in Database Summary Related Articles Check |
This paper is cited in the following contexts:
Clustered Partial Linear Regression - Torgo, Costa (2000) (Correct)
....are not obtained randomly and are usually smaller than the original training sample. The type of resampling carried out by our clustering step changes the distribution of the cases in the original sample, which is not the case with bagging. From this perspective, our method is related to boosting [16, 10], where the same distribution change is done through a system of weights. However, contrary to boosting our method is not sequential and thus it is possible to construct the individual models in parallel. Devogelaere et al. 7] describe a related approach to regression. Their GAdC system performs ....
Shapire,R. : The strength of weak learnability. Machine Learning, 5, 197-227. Kluwer Academic Publishers, 1990.
Layered Learning - Stone, Veloso (1999) (10 citations) (Correct)
....repeated as champion in a eld of 37 teams. Full details of the competitions are available at www.robocup.org. 6. Related Work The original hierarchical learning constructs were devised to improve the generalization of a single learning task by running multiple learning processes. Both boosting (Shapire, 1990) and stacked generalization (Wolpert, 1992) improve function generalization by combining the results of several generalizers or several runs of the same generalizer. These approaches contrast with layered learning in that the layers in layered learning each deal with di erent tasks. Boosting or ....
Shapire, R. E. (1990). The strength of weak learnability.
Actively Searching for an Effective Neural-Network Ensemble - Opitz, Shavlik (1996) (26 citations) (Correct)
....training set (Breiman, 1996a; Krogh Vedelsby, 1995) Unlike Addemup however, these approaches do not directly address how to generate such networks that are optimized for the ensemble as a whole. One method that does actively create members for its ensemble, however, is the Boosting algorithm (Shapire, 1990). Boosting converts any learner that is guaranteed to always perform slightly better than random guessing into one that achieves arbitrarily high accuracy. Drucker et al. 1992) applied Boosting to neural networks to improve their error rate on a handwrittendigit recognition task. A problem with ....
Shapire, R. (1990). The strength of weak learnability. Machine Learning, 5:197--227.
A Quantum Computational Learning Algorithm - Ventura, Martinez (1998) (Correct)
....combine them in such a way as to produce a good hypothesis for the function. Jackson s learning algorithm, which he calls the Harmonic Sieve, combines techniques from discrete Fourier analysis due to Goldreich and Levin [Gol89] and Kushilevitz and Mansour [Kus93] with the boosting ideas of Shapire [Sha90] and Freund [Fre90] Fre92] The Harmonic Sieve guarantees strong learnability of DNF with two caveats: first, the function distribution D must be uniform (actually this is relaxed somewhat, but it is still restrictive) and second, access to a membership oracle M rather than to an example oracle ....
Shapire, R. E., "The Strength of Weak Learnability", Machine Learning, vol. 5, pp. 197-227, 1990.
An Experimental Evaluation of Coevolutive Concept Learning - Anglano, Giordana, Bello, .. (1998) (12 citations) (Correct)
....described here, conforms to the one proposed by (Potter et al. 1995) properly re interpreted in the framework of concept learning, which naturally conforms to it. Finally, the reassignment of the examples to be covered to G nodes, performed by the Supervisor, can be considered a kind of boosting (Shapire, 1990): in subsequent runs, the search efforts shall be concentrated on those parts of the hypothesis space not yet adequately covered. Currently, the series of found hypotheses are combined into a unique formula, which differentiate this approach from a genuine boosting. However, nothing hinders the ....
Shapire, R. (1990). The strength of weak learnability.
Machine Learning (Lecture notes 13) - Rivest (1994) Self-citation (Shapire) (Correct)
....26, 1994 Lecturer: Ron Rivest Scribe: Rob Miller Outline ffl Finish the proof that weak learnability implies strong learnability. 13.1 Weak Learnability Implies Strong Learnability Today we finish the proof that weak learnability implies strong learnability, a theorem due originally to Shapire [3]. This is a rather surprising result; essentially it says that if we can find an efficient learning algorithm that guarantees only a fixed accuracy ffl 0 (ffl 0 1 2 Gamma 1 p(n) with only a fixed confidence 1 Gamma ffi 0 , then we can turn it into a general PAC algorithm for arbitrary ffl ....
R.E. Shapire. The strength of weak learnability. Machine Learning, 5(2):197-227, 1990.
Ensemble Learning with Biased Classifiers: The Triskel.. - Heß, Khoussainov.. (2005) (Correct)
No context found.
Shapire, R.E.: The Strength of Weak Learnability. Machine Learning 5 (1990) 197--227
Unknown - Lazy Kernel-Density-Based.. (Correct)
No context found.
R. Shapire, "The Strength of Weak Learnability," Machine Learning, Vol. 5 No.6, pp. 197-227, 1990.
Ensemble of Artificial Neural Networks for.. - Silva, de.. (Correct)
No context found.
Shapire, R.E. The strength of weak learnability. Machine Learning, 5(2):197-227, 1990.
A Genetic Algorithm Approach for Creating Neural-Network.. - Opitz, Shavlik (1999) (2 citations) (Correct)
No context found.
R. Shapire. The strength of weak learnability. Machine Learning, 5:197--
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC