Jenq-Neng Hwang, Jai J. Choi, Seho Oh & Robert J. Marks II. QueryBased learning applied to Partially Trained Multilayer Perceptrons. IEEE Trans. on Neural Networks, 2(1), p. 131-136, janvier 1991. Home/Search Document Not in Database Summary Related Articles Check |
This paper is cited in the following contexts:
Apprentissage Dans Les Réseaux Récurrents Pour La Modélisation.. - Szilas (1995) (Correct)
....de celles reprsentes sur la figure III.4. actif passif Erreur Temps Figure III.4. Courbes d apprentissages dans le cas passif et dans le cas actif En corollaire, slectionner les exemples les plus informatifs permet de diminuer l erreur finale de la solution obtenue [Sanzogni Vaccaro 93] Hwang et al. 91] En effet, sans prolonger l apprentissage excessivement, l apprentissage actif converge plus vite vers la solution, et donne donc une erreur plus faible (voir figure III.4) Deuximement, slectionner les exemples les plus informatifs peut aider trouver une meilleure solution du point de vue de ....
....slection active permet par exemple de corriger une base d apprentissage dont les exemples ne sont pas rpartis uniformment. 164 Troisimement, dans certaines applications, obtenir un exemple d apprentissage peut tre assez coteux, quand un procd de mesure complexe est mis en jeu [Atlas et al. 90] Hwang et al. 91] RayChaudhuri Hamey 95] Dans ce cas, on aura intrt chantillonner l environnement le moins possible et donc de choisir chaque fois l exemple le plus utile. Quatrimement, on peut envisager l apprentissage actif informatif comme le moyen de sortir d un minimum local#: une fois le rseau ....
[Article contains additional citation context not shown here]
Jenq-Neng Hwang, Jai J. Choi, Seho Oh & Robert J. Marks II. QueryBased learning applied to Partially Trained Multilayer Perceptrons. IEEE Trans. on Neural Networks, 2(1), p. 131-136, janvier 1991.
Heterogeneous Uncertainty Sampling for Supervised Learning - Lewis, Catlett (1994) (70 citations) (Correct)
....that one classifier is not representative of the set of all classifiers consistent with the labeled data: the version space [24] The degree to which this is a problem in practice has not been established. Single classifier approaches have successfully been used in generating arbitrary queries [16] and in sampling from labeled data [8, 25] Uncertainty sampling with a single classifier can also be viewed as a variation on the heuristic of training on misclassified instances [15, 33, 35] A familiar example of this is windowing, which appeared in Quinlan s first paper on ID3 [26] was ....
....as the two instances with P (Cjw) s closest to, but below 0.5. Using a subsample size of four rather than one was a compromise for efficiency. Selecting examples both above and below 0. 5 was a simple way to halve the potential number of duplicate examples, and may also have benefits for training [16]. 5.2 Initial Classifier Without an initial classifier our sampling algorithm would commence with a long period of nearly random sampling before finding any examples of a low frequency class. Obtaining a plausible initial classifier is usually easy it would be surprising if an expert were able ....
Jenq-Neng Hwang, Jai J. Choi, Seho Oh, and Robert J. Marks II. Query-based learning applied to partially trained multilayer perceptrons. IEEE Transactions on Neural Networks, 2(1):131--136, January 1991.
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