作者: Gasser Auda , Mohamed Kamel
关键词:
摘要: There is a wide variety of Modular Neural Network (MNN) classifiers in the literature. They differ according to design their architecture, task-decomposition scheme, learning procedure, and multi-module decision-making strategy. Meanwhile, there lack comparative studies MNN This paper compares ten which give good representation varieties, viz., Decoupled; Other-output; ART-BP; Hierarchical; Multiple-experts; Ensemble (majority vote); (average Merge-glue; Hierarchical Competitive Net; Cooperative Net. Two benchmark applications different degree nature complexity are used for performance comparison, strength-points drawbacks networks outlined. The aim help potential user choose an appropriate model application hand available computational resources.