| Document Date: 2009-12-17 20:19:46
Open Document File Size: 1,84 MBShare Result on Facebook
Company IEEE Authorized / A. Proposed Solutions / Authorized / Agilent Technologies / / Country Jordan / / Currency USD / / / Facility Agilent Technologies High Frequency Technology Center / Oklahoma State University / / IndustryTerm parameter learning algorithm / simplest possible recurrent network / basic gradient-descent algorithm / gradient-descent algorithm / gradient search algorithm falls / neural network / recurrent neural networks / certain recurrent networks / recurrent networks / communication systems / to arbitrary recurrent network / neural networks / recurrent network / unstable network / line search / nonlinear recurrent network / batch training algorithms / linear network / nonlinear networks / linear recurrent network / firstorder network / backpropagation algorithm / sequence processing / line search minimization / larger networks / linear recurrent networks / power systems / multilayer feedforward networks / search direction / simplest network / nonlinear network / complicated networks / large nonlinear neural networks / golden section line search / / NaturalFeature Spurious Valleys / / Organization US Federal Reserve / Oklahoma State University / School of Electrical and Computer Engineering / / Person Chen / Martin T. Hagan / Wang / / Position neural-network-based model reference controller / / Technology speech recognition / neural network / training algorithm / batch training algorithms / gradient-descent algorithm / quasi-Newton algorithm / parameter learning algorithm / simulation / backpropagation algorithm / Digital Object Identifier / basic gradient-descent algorithm / / URL http /
SocialTag |
686 IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. 20, NO. 4, APRIL 2009 Spurious Valleys in the Error Surface of Recurrent Networks—Analysis and Avoidance