123 search results - page 12 / 25 » Multi-information Ensemble Diversity |
ICPR
2004
IEEE
14 years 9 months ago
2004
IEEE
Recent empirical work has shown that combining predictors can lead to significant reduction in generalization error. The individual predictors (weak learners) can be very simple, ...
CEC
2009
IEEE
13 years 11 months ago
2009
IEEE
—When performing predictive data mining, the use of ensembles is known to increase prediction accuracy, compared to single models. To obtain this higher accuracy, ensembles shoul...
ICDAR
2011
IEEE
12 years 8 months ago
2011
IEEE
—Handwritten text line segmentation on real-world data presents significant challenges that cannot be overcome by any single technique. Given the diversity of approaches and the...
IJCAI
2003
13 years 9 months ago
2003
Ensemble methods like bagging and boosting that combine the decisions of multiple hypotheses are some of the strongest existing machine learning methods. The diversity of the memb...
MCS
2004
Springer
14 years 1 months ago
2004
Springer
Abstract. One of the potential advantages of multiple classifier systems is an increased robustness to noise and other imperfections in data. Previous experiments on classificati...