Ensemble deep learning: A review

Ensemble learning combines several individual models to obtain better generalization performance. Currently, deep architectures are showing performance compared the shallow or traditional models. Deep ensemble combine advantages of both as well such that final model has This paper reviews state-of-art and hence serves an extensive summary for researchers. The broadly categorized into bagging, boosting, stacking, negative correlation based models, explicit/implicit ensembles, homogeneous/heterogeneous ensemble, decision fusion strategies Applications in different domains also briefly discussed. Finally, we conclude this with some potential future research directions. • categorizes discussed boosting strategies, unsupervised, semi supervised, reinforcement online/incremental, multilabel Application provides outlook towards