请问有对神经网络和经济预测感兴趣的学友吗？

想学，趁现在工作还有点联系，多接触接触。

神经网络设计(汉)c.pdf (5.27 MB, 下载次数: 9, 售价: 26 个论坛币) ，怎么一个论坛币也没入账，依然是427枚论坛币，坛主啊？出什么事情了？冤枉啊！

regressi0n 发表于 2011-12-29 23:56
神经网络设计(汉)c.pdf (5.27 MB, 下载次数: 9, 售价: 26 个论坛币) ，怎么一个论坛币也没入账，依然是427枚 ...

上传神经网络方面的宝贝被下载那么多次都不挣坛币，真不知道要是上传了神经网络视频会不会也......

下边这段程序哪里写得不合适,warning+error那么疯狂？warning很多，有说“TRAINWH函数系已废弃不用的函数，用NNT2LIN和TRAIN升级并训练网络，键入NNTWARN OFF关闭NNT报警信息（TRAINWH is an obsolete function. Use NNT2LIN and TRAIN to update and train your network. Type NNTWARN OFF to suppress NNT warning messages）”的，有说“PURELIN用法过时，请选用PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B)，键入NNTWARN OFF关闭NNT报警信息（PURELIN used in an obsolete way. Use PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B). Type NNTWARN OFF to suppress NNT warning messages.）”，......，凡此种种。请问哪位坛友可以帮忙用不过时的语句更正在下的拙作？谢谢！

[题目]设计自适应线性网络实现从输入矢量到输出适量的变换关系。其输入矢量和输出适量分别为：

P=[1.0 -1.2]
T=[0.5 1.0]

解：
用自适应线性网络求解问题时，设计者要确定期望误差值，以及最大循环次数。对此提可分别选err_goal=0.001; max_epoch=20，自适应线性网络设计程序可写为：

%wf1.m
%
P=[1.0 -1.2]
T=[0.5 1.0]
[R,Q]=size(P);
[S,Q]=size(T);
[W,B]=rands(S,R);
max_epoch=20; %最大循环次数
err_goal=0.001; %期望误差
lr=0.4*maxlinlr(P); %最佳学习速率
disp_freq=1;  %设置显示频率
TP=[disp_freq max_epoch err_goal lr];  %设置参数变量TP
[W,B,epochs,error]=trainwh(W,B,P,T,TP) %进行线性网络权值训练
end
P =
    1.0000   -1.2000

T =
    0.5000    1.0000

*WARNING* TRAINWH is an obsolete function.
          Use NNT2LIN and TRAIN to update and train your network.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* TRAINWB is an obsolete function.
          Use TRAINB to train your network.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* PURELIN used in an obsolete way.
          Use PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B).
          Type NNTWARN OFF to suppress NNT warning messages.

TRAINWH: 0/20 epochs, SSE = 2.65292.

*WARNING* BARERR is an obsolete function.
          Use BAR to make bar plots.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* LEARNP used in an obsolete way.
          See help on LEARNP for new argument list.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* PURELIN used in an obsolete way.
          Use PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B).
          Type NNTWARN OFF to suppress NNT warning messages.

TRAINWH: 1/20 epochs, SSE = 0.922534.

*WARNING* BARERR is an obsolete function.
          Use BAR to make bar plots.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* LEARNP used in an obsolete way.
          See help on LEARNP for new argument list.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* PURELIN used in an obsolete way.
          Use PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B).
          Type NNTWARN OFF to suppress NNT warning messages.

TRAINWH: 2/20 epochs, SSE = 0.321687.

*WARNING* BARERR is an obsolete function.
          Use BAR to make bar plots.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* LEARNP used in an obsolete way.
          See help on LEARNP for new argument list.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* PURELIN used in an obsolete way.
          Use PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B).
          Type NNTWARN OFF to suppress NNT warning messages.

TRAINWH: 3/20 epochs, SSE = 0.112583.

*WARNING* BARERR is an obsolete function.
          Use BAR to make bar plots.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* LEARNP used in an obsolete way.
          See help on LEARNP for new argument list.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* PURELIN used in an obsolete way.
          Use PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B).
          Type NNTWARN OFF to suppress NNT warning messages.

TRAINWH: 4/20 epochs, SSE = 0.0395919.

*WARNING* BARERR is an obsolete function.
          Use BAR to make bar plots.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* LEARNP used in an obsolete way.
          See help on LEARNP for new argument list.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* PURELIN used in an obsolete way.
          Use PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B).
          Type NNTWARN OFF to suppress NNT warning messages.

TRAINWH: 5/20 epochs, SSE = 0.0140114.

*WARNING* BARERR is an obsolete function.
          Use BAR to make bar plots.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* LEARNP used in an obsolete way.
          See help on LEARNP for new argument list.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* PURELIN used in an obsolete way.
          Use PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B).
          Type NNTWARN OFF to suppress NNT warning messages.

TRAINWH: 6/20 epochs, SSE = 0.00499911.

*WARNING* BARERR is an obsolete function.
          Use BAR to make bar plots.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* LEARNP used in an obsolete way.
          See help on LEARNP for new argument list.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* PURELIN used in an obsolete way.
          Use PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B).
          Type NNTWARN OFF to suppress NNT warning messages.

TRAINWH: 7/20 epochs, SSE = 0.00180216.

*WARNING* BARERR is an obsolete function.
          Use BAR to make bar plots.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* LEARNP used in an obsolete way.
          See help on LEARNP for new argument list.
          Type NNTWARN OFF to suppress NNT warning messages.


*WARNING* PURELIN used in an obsolete way.
          Use PURELIN(NETSUM(Z,B*ones(1,Q))) instead of PURELIN(Z,B).
          Type NNTWARN OFF to suppress NNT warning messages.

TRAINWH: 8/20 epochs, SSE = 0.000658081.

*WARNING* BARERR is an obsolete function.
          Use BAR to make bar plots.
          Type NNTWARN OFF to suppress NNT warning messages.

W =
   -0.2108

B =
    0.7300

epochs =
     8

error =
  Columns 1 through 8
    2.6529    0.9225    0.3217    0.1126    0.0396    0.0140    0.0050    0.0018
  Column 9
    0.0007
??? end
    |
Error: Illegal use of reserved keyword "end".

regressi0n 发表于 2011-12-26 02:45
1 真不容易，化整为零，终于全部上传成功。这个坛子上转文件怎么这么费劲？！
2 定价有些蹊跷， ...

老乡们，大家上传文件挣论坛币一定要小心，这个坛子肯定出问题了，上传的东东被下载了n多次，就是看不到论坛币余额增加！看不到就看不到吧，似乎还没有人管！！

taige5555 发表于 2011-12-26 20:13
想学，趁现在工作还有点联系，多接触接触。

多接触很好，那就请您帮忙解决几个小问题吧。具体问题见本人的帖子。

下边这段程序里写“net.trainParam.show = 50”的作用应该是显示：
               TRAINGD, Epoch 0/300, MSE 1.59423/1e-05, Gradient 2.76799/1e-10
               TRAINGD, Epoch 50/300, MSE 0.00236382/1e-05, Gradient 0.0495292/1e-10
               TRAINGD, Epoch 100/300, MSE 0.000435947/1e-05, Gradient 0.0161202/1e-10
               TRAINGD, Epoch 150/300, MSE 8.68462e-05/1e-05, Gradient 0.00769588/1e-10
               TRAINGD, Epoch 200/300, MSE 1.45042e-05/1e-05, Gradient 0.00325667/1e-10
               TRAINGD, Epoch 211/300, MSE 9.64816e-06/1e-05, Gradient 0.00266775/1e-10
               TRAINGD, Performance goal met.
但是，程序运行的结果怎么变了味儿？有能做出解释的高人吗？谢谢！

>> p = [-1 -1 2 2;0 5 0 5];
t = [-1 -1 1 1];
>>
>> net=newff(minmax(p),[3,1],{'tansig','purelin'},'traingd');
>> net.trainParam.show = 50;
net.trainParam.lr = 0.05;
net.trainParam.epochs = 300;
net.trainParam.goal = 1e-5

net =

    Neural Network object:

    architecture:

         numInputs: 1
         numLayers: 2
       biasConnect: [1; 1]
      inputConnect: [1; 0]
      layerConnect: [0 0; 1 0]
     outputConnect: [0 1]
     targetConnect: [0 1]

        numOutputs: 1  (read-only)
        numTargets: 1  (read-only)
    numInputDelays: 0  (read-only)
    numLayerDelays: 0  (read-only)

    subobject structures:

            inputs: {1x1 cell} of inputs
            layers: {2x1 cell} of layers
           outputs: {1x2 cell} containing 1 output
           targets: {1x2 cell} containing 1 target
            biases: {2x1 cell} containing 2 biases
      inputWeights: {2x1 cell} containing 1 input weight
      layerWeights: {2x2 cell} containing 1 layer weight

    functions:

          adaptFcn: 'trains'
           initFcn: 'initlay'
        performFcn: 'mse'
          trainFcn: 'traingd'

    parameters:

        adaptParam: .passes
         initParam: (none)
      performParam: (none)
        trainParam: .epochs, .goal, .lr, .max_fail,
                    .min_grad, .show, .time

    weight and bias values:

                IW: {2x1 cell} containing 1 input weight matrix
                LW: {2x2 cell} containing 1 layer weight matrix
                 b: {2x1 cell} containing 2 bias vectors

    other:

          userdata: (user stuff)

zhanfengsky 发表于 2011-10-27 14:51
正在准备学习……

这张照片好可爱、好可爱，真是名副其实的初中生，哈哈！

regressi0n 发表于 2012-1-3 03:06
下边这段程序里写“net.trainParam.show = 50”的作用应该是显示：
               TRAINGD, Epoch 0/300, ...

还在吗？报错到底是不是因为版本太低啊

regressi0n 发表于 2012-1-3 03:06
下边这段程序里写“net.trainParam.show = 50”的作用应该是显示：
               TRAINGD, Epoch 0/300, ...

出现了同样地结果