Hybrid negative correlation learning with randomly splitting data

This paper discusses how to apply the ensemble learning for the individual learners on the randomly splitting data. Rather than letting the individual learners learn independently on the different subsets, it would be better for the individual learners to learn cooperatively by exchanging the learned values. In this way, the individual learners could learn the whole given data together while they can learn to be different at the same time. Negative correlation learning with splitting data is implemented, and compared to other independent learning implementations. The results suggest that the randomly splitting data could let the ensemble system learn better and more stable on both the training data and the testing data. One surprising result on the ensemble system with the simple individual learners is that the ensemble system had learned better when the individual learners were trained on the less splitting data.