Echo State Network with Hub Property

摘要：

　　Reservoir computing,as an effective approach of training recurrent neural network,has drawn wide attention for its superior performance on some benchmark problems.Reservoirs which are closer to the topology of the brain have shown better performance.It has been found that some neurons in the brain act as hubs integrating and distributing information.By simulating the hub property,this paper proposed a novel topology for dynamical reservoir,called echo state network(ESN)with hub property(HESN).Different from conventional reservoir,some neurons acting as hubs are added to the reservoir of HESN.Experimental results show that our proposed approach has improved the performance of the original ESN on two benchmark datasets.

关键词： Reservoir computing Echo state network Hub Prediction

作者: Fanjun Li Ying Li Xiaohong Wang

作者单位: School of Mathematical Sciences,University of Jinan,Jinan,China School of Science,Qilu University of Technology,Jinan,China School of Electrical Engineering,University of Jinan,Jinan,China

会议类型: 国际会议

会议名称: 2019中国智能自动化大会(CIA,2019)

会议地点: 江苏镇江

会议语种:英文

页码: 537-544

在线出版日期: 2019-09-20（万方平台首次上网日期，不代表论文的发表时间）