Soft Computing-Neural Networks Ensembles

Neural Network ensemble is a learning paradigm where a collection of finite number of neural networks is trained for the same task. It is understood that the generalization ability of neural networks, i.e., training many neural networks and then combining their predictions. ANN ensemble techniques have become very popular amongst neural network practitioners in a variety of ANN application domains. There are many different ensemble techniques, but the most popular include some elaboration of bagging and boosting or stacking. When applied to Ann's, ensemble techniques can produce dramatic improvements in generalization performance.