Issues in Stacked Generalization Export

@article{citeulike:4077506, abstract = {Stacked generalization is a general method of using a high-level model to combine lowerlevel models to achieve greater predictive accuracy. In this paper we address two crucial issues which have been considered to be a `black art\&\#039; in classification tasks ever since the introduction of stacked generalization in 1992 by Wolpert: the type of generalizer that is suitable to derive the higher-level model, and the kind of attributes that should be used as its input. We find that best results are obtained when the higher-level model combines the confidence (and not just the predictions) of the lower-level ones. We demonstrate the effectiveness of stacked generalization for combining three different types of learning algorithms for classification tasks. We also compare the performance of stacked generalization with majority vote and published results of arcing and bagging.}, author = {Ting, Kai M. and Witten, Ian H.}, citeulike-article-id = {4077506}, citeulike-linkout-0 = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.34.948}, journal = {Journal of Artificial Intelligence Research}, keywords = {confidence, ensemble, probability, stacking}, pages = {271--289}, posted-at = {2009-08-13 06:09:51}, priority = {1}, title = {Issues in Stacked Generalization}, url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.34.948}, volume = {10}, year = {1999} }

TY - JOUR ID - citeulike:4077506 L3 - citeulike-article-id:4077506 N2 - Stacked generalization is a general method of using a high-level model to combine lowerlevel models to achieve greater predictive accuracy. In this paper we address two crucial issues which have been considered to be a `black art' in classification tasks ever since the introduction of stacked generalization in 1992 by Wolpert: the type of generalizer that is suitable to derive the higher-level model, and the kind of attributes that should be used as its input. We find that best results are obtained when the higher-level model combines the confidence (and not just the predictions) of the lower-level ones. We demonstrate the effectiveness of stacked generalization for combining three different types of learning algorithms for classification tasks. We also compare the performance of stacked generalization with majority vote and published results of arcing and bagging. JF - Journal of Artificial Intelligence Research EP - 289 TI - Issues in Stacked Generalization VL - 10 SP - 271 KW - confidence KW - ensemble KW - probability KW - stacking AU - Ting, Kai AU - Witten, Ian PY - 1999/// UR - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.34.948 ER -