Back to list of seminars for 2006/07
Previous work on evolving neural networks has focused on single neural networks. However, monolithic neural networks become too complex to train and evolve for large and complex problems. It is often better to design a collection of simpler neural networks that work collectively and cooperatively to solve a large and complex problem. The key issue here is how to design such a collection automatically so that it has the best generalisation. This talk introduces some recent work on evolving neural network ensembles, including negative correlation, constructive negative correlation and multi-objective approaches to ensemble learning.
On negative correlation: Y. Liu and X. Yao, ``Simultaneous training of negatively correlated neural networks in an ensemble,' IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, 29(6):716-725, December 1999.
On negative correlation and evolution: Y. Liu, X. Yao and T. Higuchi, ``Evolutionary Ensembles with Negative Correlation Learning,' IEEE Transactions on Evolutionary Computation, 4(4):380-387, November 2000.
On constructive ensemble algorithms: Md. Monirul Islam, X. Yao and K. Murase, ``A constructive algorithm for training cooperative neural network ensembles,' IEEE Transactions on Neural Networks, 14(4):820-834, July 2003.
On multi-objective approaches to ensemble learning: A Chandra and X. Yao, ``Ensemble learning using multi-objective evolutionary algorithms,' Journal of Mathematical Modelling and Algorithms, 5(4):417-445, December 2006.