Methoden vergleichen

	Ensemble Support Vector Machine (Ensemble-Stützvektor-Maschine)	Boosting	Stacking
Fachgebiet	Maschinelles Lernen	Maschinelles Lernen	Maschinelles Lernen
Familie	Machine learning	Machine learning	Machine learning
Entstehungsjahr	2000–2003	1990–1997	1992
Urheber	Kim, H.-C. et al.; Dietterich, T. G.	Schapire, R. E.; Freund, Y.	Wolpert, D.H.
Typ	Ensemble of SVMs (bagging, voting, or stacking)	Sequential ensemble (iterative reweighting)	Ensemble (heterogeneous meta-learning)
Wegweisende Quelle	Kim, H.-C., Pang, S., Je, H.-M., Kim, D., & Bang, S. Y. (2002). Constructing support vector machine ensemble. Pattern Recognition, 36(12), 2757–2767.	Freund, Y. & Schapire, R. E. (1997). A decision-theoretic generalization of on-line learning and an application to boosting. Journal of Computer and System Sciences, 55(1), 119–139.	Wolpert, D.H. (1992). Stacked Generalization. Neural Networks, 5(2), 241–259.
Aliasnamen	Ensemble SVM, SVM ensemble, bagged SVM, SVM committee machine	AdaBoost, gradient boosting, iterative reweighting ensemble, sequential ensemble	Stacking (Yığınlama — Meta-Öğrenme), stacked generalization, meta-learning ensemble, super learner
Verwandt	5	6	5
Zusammenfassung	Ensemble Support Vector Machine combines multiple independently trained SVM classifiers or regressors — each fitted on a different data partition, bootstrap sample, or feature subset — and aggregates their outputs via voting, averaging, or stacking. The approach mitigates the high computational cost and sensitivity to kernel hyperparameters inherent in a single large-scale SVM, while improving generalisation on complex or high-dimensional datasets.	Boosting is a sequential ensemble technique that converts many simple, barely-better-than-chance learners into a single highly accurate model by repeatedly focusing training on the examples that previous learners got wrong, then combining all learners with weights proportional to their individual accuracy.	Stacking, or stacked generalization, is an ensemble method introduced by David Wolpert in 1992 that combines the outputs of several different base models (Level-0) through a separate meta-model (Level-1). Unlike bagging and boosting, it deliberately uses heterogeneous model types, and it is the standard final-stage strategy in Kaggle competitions.