Intelligent Object and Pattern Recognition using Ensembles in Back Propagation Neural Network
Hafiz T. Hassan, Muhammad Usman Khalid, Kashif Imran · 2010
Abstract — Object and Pattern Recognition applications are one of the most active research fields in the Neural Network research community today. Neural networks are mostly preferred in predictive and classification problems because they are unstable classifier that is little change in input causes drastic changes in output. Now in some scenarios the Neural Networks are not enough to do the required job such as classification or regression analysis so an evolved architecture called ensembles is used that is instead of one classifier such as a neural network many are used. In the paper we are going to incorporate Back Propagation Neural Network for the classification of two datasets (the sonar and ionosphere datasets) in bagging ensemble architecture using PCA and standalone architecture.