建立神经网络输入参数过多怎么办

yuchaowangok

>  net=newff(p',t,[300,25,1],['tansig','tansig','purelim'],'trainrp','mse');
# E* F  F( U! w1 {5 W??? Cell contents reference from a non-cell array object.

$ u# T4 Z6 v9 L7 Y* w: I2 IError in ==> newff>new_5p1 at 199
" z* B. t( P1 a1 S, R0 r; A% o* f  if (length(tf) < i) || all(isnan(tf{i}))

3 ]/ Z, y* U, l7 ]0 G% nError in ==> newff>create_network at 130
( m! V! q" a2 g: ^, q& f# m7 C  net = new_5p1(varargin{:});
4 \7 n: y( J; @6 u2 F4 Y% I3 ^
  \# _% }1 z) f% @2 }9 a2 c" nError in ==> newff at 102
0 _4 E! t4 ~2 f% N2 u4 f) H  out1 = create_network(varargin{:});

yuchaowangok

我的这个网络输入参数有30个，但输入元素个数是300个

darker50

   一般就是降维，让参数变少呗！