A SELF-GENERATING MODULAR NEURAL-NETWORK ARCHITECTURE FOR SUPERVISED LEARNING

In this paper, we present a self-generating modular neural network arc hitecture for supervised learning. In the architecture, any kind of fe edforward neural networks can be employed as component nets. For a giv en task, a tree-structured modular neural network is automatically gen erated with a growing algorithm by partitioning input space recursivel y to avoid the problem of pre-determined structure. Due to the princip le of divide-and-conquer used in the proposed architecture, the modula r neural network can yield both good performance and significantly fas ter training. The proposed architecture has been applied to several su pervised learning tasks and has achieved satisfactory results.