神经网络极速学习方法研究_邓万宇.pdfVIP

33 2 V ol . 33 No . 2 2010 2 CH IN ESE JO U RN A L O F CO MP UT ERS F eb. 20 10 1) 1) 2) 1) 邓万宇 郑庆华 陈 琳 许学斌 1) ( 7 10049) 2) ( 7 1006 1) ( Single-hidden L ayer F eedf o rw ard Neura l N etw o rk , SL F N) , , . : ( 1) ( Back Pro pag atio n, BP ) , ( 2) . . , E LM ( R EL M ) , , . R EL M E LM BP SV M . T P 18 DOI : 10. 37 24/ SP. J. 10 16 . 2010. 00279 Research on Extreme Learning of Neural Networks 1) 1) 2) 1) DEN G Wan- Y u ZH EN G Q ing- H u a CH EN Lin XU Xu e- Bin 1) (Ministry of Education K ey Laboratory f or Intelligent N etworks and N etwork Security, Dep art ent of Co p uter Science and Technology, Xican Jiaotong University, Xican 7 10049) 2) (Dep art entof Co p uter Science and Technology , Xican University of Posts Teleco unications, X ican 7 10 12 1) Abstract SLF N s( Single-hidden L ayer F eed fo rw ar d N eur al net w ork s) have been w idely applied in many fields inclu ding p att ern recog nit ion, aut om at ic contr ol, dat a mining et c. H ow ever, t he t radit ional lear ning methods can not m eet t he act ual needs due t o t w o m ain reaso ns . F irst ly, t he t radit ional met hod is m