Using Negative Correlated Networks to Improve Neural Ensemble Performance

This paper investigates the effect of diversity caused by Negative Correlation Learning(NCL) in the combination of neural classifiers and presents an efficient way to improve combining performance. Decision Templates and Averaging, as two non-trainable combining methods and Stacked Generalization as a trainable combiner are investigated in our experiments. Utilizing NCL for diversifying the base classifiers leads to significantly better results in all employed combining methods. Experimental results on five datasets from UCI and ELENA repositories indicate that by employing NCL, the performance of the ensemble structure can be more favorable compared to that of an ensemble use independent base classifiers.