网格环境下分布式元学习任务的设计

元学习方法是采用集成学习的方式来生成最终的全局预测模型。该方法的基本思想是从已经获得的知识中再进行学习,从而得到最终的数据模式。网格能有效地为元学习提供高性能和分布式的基础设施。文中根据知识网格的概念,在Globus Toolkit的基础上,分析了知识网格的体系结构和它的主要组件。根据分布式元学习的一般过程,设计了在知识网格体系结构下的元学习任务。

Design of Distributed Meta-Learning Task on Grid

Meta-learning method makes use of ensemble learning to obtain the global predication module.Its primary idea is to learn again according to discovered knowledge,finally to get data patterns.The grid can be effectively exploited for deploying meta-learning because of the high-performance it can offer and its distributed infrastructure.According to the concept of knowledge grid,analyzes the knowledge grid architecture and its main components on top of the Globus Toolkit,designs a kind of software modules according to the process of data mining,and presents the modules should provide services.Presents the design of meta-learning task of the grid on basis of the process of distributed meta-learning.