The code scheme of network structure,and the corresponding genetic operators are designed. As a result,the algorithm converges on the overall optimal structure.

• 针对这种分类器的结构学习,设计了结合对数似然的适应度函数及相应的遗传算子,并给出了网络结构的编码方案,使得该算法能够收敛到全局最优的结构。