kevin19821 发表于 2007-10-28 14:54

newrbe这个命令是作啥用的呀?

哪位高人给我讲解一下呀

[ 本帖最后由 eight 于 2007-10-28 16:29 编辑 ]

eight 发表于 2007-10-28 16:29

原帖由 kevin19821 于 2007-10-28 14:54 发表 http://www.chinavib.com/forum/images/common/back.gif
哪位高人给我讲解一下呀


>> help newrbe
NEWRBE Design an exact radial basis network.

   Synopsis

   net = newrbe(P,T,SPREAD)

   Description

   Radial basis networks can be used to approximate functions.
   NEWRBE very quickly designs a radial basis network with
   zero error on the design vectors.

   NEWRBE(P,T,SPREAD) takes two or three arguments,
   P      - RxQ matrix of Q input vectors.
   T      - SxQ matrix of Q target class vectors.
   SPREAD - of radial basis functions, default = 1.0.
   and returns a new exact radial basis network.

   The larger that SPREAD, is the smoother the function approximation
   will be. Too large a spread can cause numerical problems.

   Examples

   Here we design a radial basis network, given inputs P
   and targets T.

       P = ;
       T = ;
       net = newrbe(P,T);

   Here the network is simulated for a new input.

       P = 1.5;
       Y = sim(net,P)

   Algorithm

   NEWRBE creates a two layer network. The first layer has RADBAS
   neurons, and calculates its weighted inputs with DIST, and its
   net input with NETPROD.The second layer has PURELIN neurons,
   and calculates its weighted input with DOTPROD and its net inputs
   with NETSUM. Both layer's have biases.

   NEWRBE sets the first layer weights to P', and the first
   layer biases are all set to 0.8326/SPREAD, resulting in
   radial basis functions that cross 0.5 at weighted inputs
   of +/- SPREAD.

   The second layer weights IW{2,1} and biases b{2} are found by
   simulating the first layer outputs A{1}, and then solving the
   following linear expression:

          * = T

   See also sim, newrb, newgrnn, newpnn.

    Reference page in Help browser
       doc newrbe
页: [1]
查看完整版本: newrbe这个命令是作啥用的呀?