Ant colony algorithm based selective neural network ensemble

作者：Chen, Chuan^*; Chen, Youqing

来源：Dynamics of Continuous Discrete and Impulsive Systems: Series B; Applications and Algorithms , 2006, 13: 312-315.

Neural network ensemble can improve the generation ability of a single neural network. This paper presents a new selective ensemble method for regression problems based on the ant colony algorithm named ANTSEN. The ant colony algorithm has so good combination optimization ability that we use it to choose accurate and diverse component neural networks. Compared with Bagging and GASEN methods, ANTSEN has advantage both on simple average combination and weighted average combination methods. Experiments show that ANTSEN makes a good balance between accuracy and diversity for the ensemble's better performance.

• 出版日期2006-12
• 单位中山大学