Radial Basis Function Neural Network Tool Box
in [Matlab]
|
Prev: show to save a matlab generated image to specific location in GIF format
Next: Fast access to rows and columns of sparse matrices
From: Azwan on 23 May 2010 03:53 Hi Greg and other experts I have been trying to understand and use the NEWRB for RBF Neural Network. Previously I also has asked several questions about NEWRB on this Matlab Central. I'm still confused about NEWRB. I have several questions for you and hope that you have time and willing to share your experience and experts about it. 1) How is the selection of RBF center in NEWRB is done? 2) How do we estimate the spread parameters or weights for the Gaussian function in the NEWRB? Many thanks in advance for your opinions and suggestions. Azwan
From: Azwan on 23 May 2010 03:54 Hi Greg and other experts I have been trying to understand and use the NEWRB for RBF Neural Network. Previously I also has asked several questions about NEWRB on this Matlab Central. I'm still confused about NEWRB. I have several questions for you and hope that you have time and willing to share your experience and experts about it. 1) How is the selection of RBF center in NEWRB is done? 2) How do we estimate the spread parameters or weights for the Gaussian function in the NEWRB? Many thanks in advance for your opinions and suggestions. Azwan
From: Azwan on 23 May 2010 03:54 Hi Greg and other experts I have been trying to understand and use the NEWRB for RBF Neural Network. Previously I also has asked several questions about NEWRB on this Matlab Central. I'm still confused about NEWRB. I have several questions for you and hope that you have time and willing to share your experience and experts about it. 1) How is the selection of RBF center in NEWRB is done? 2) How do we estimate the spread parameters or weights for the Gaussian function in the NEWRB? Many thanks in advance for your opinions and suggestions. Azwan
From: Greg Heath on 23 May 2010 18:40 On May 23, 3:54 am, "Azwan " <azwaniskan...(a)unimap.edu.my> wrote: > Hi Greg and other experts > > I have been trying to understand and use the NEWRB for RBF Neural > Network. Previously I also has asked several questions about NEWRB on > this Matlab Central. I'm still confused about NEWRB. I have several > questions for you and hope that you > have time and willing to share your experience and experts about it. > > 1) How is the selection of RBF center in NEWRB is done? The input chosen to become the nexr center is the one that has the smallest measure of correlation with the target as measured by the sum of the 4th powers of the input-target correlation coefficients. The command type newrb will reveal the source code. > 2) How do we estimate the spread parameters or weights for > the Gaussian function in the NEWRB? By trial and error. See my post on rbfnn design. Search on greg heath newrb rbfnn. > Many thanks in advance for your opinions and suggestions. You are perfectly welcome. Greg
From: Azwan on 23 May 2010 22:28
Hi Greg, Another question again. I'm wondering what type of basis/ transfer function that is used in newrb. I know that it uses radbas. Is this the Gaussian type of transfer function or something similar or different kind. Thanks again for your help Azwan |