MATLAB: Is it possible to connect nodes in one layer to a specific node in the next layer using the Matlab neural net toolbox

classificationneural network

Hi,

I'm completely new to Matlab and I'm finding it difficult to find any information on this particular problem. I'm trying to create a probability neural network. Im using newpnn and modifying it for my own needs. As it stands I have 20 nodes in my pattern layer. I want to connect 10 to one node and 10 to the other in the next layer (the summation layer). Is this possible? I have attached a diagram to show what I mean.

Also how would I perform kernel density estimation in the summation nodes?

Thanks,

Steve

Best Answer

The PNN has a single hidden layer. The output layer AUTOMATICALLY performs the summation.

Why would you want to make the net more complicated by adding a separate summation layer?

The only thing you have to do is

1. Make sure your target matrix only contains columns from eye(2).

2. Divide the data into trn/val/tst subsets.

3. Standardize the training data and use the mean and stdv vectors to normalize the val and test data.

4.Initialize the best val subset error rate bestspread = 0, bestvalerr = inf

5. Loop over positive spread values surrounding the default of 1 (e.g., 0.1:0.1:2)

    a.Use the training data to create a net
    b. Obtain and store the train, val and test subset error rates vs spread
    c. Compare the val error rate with the best. If is lower, it becomes the best.

6. The test set error rate corresponding to the minimum val set error is the unbiased estimate of error on unseen non-training data.

7. If performance is unsatisfactory consider a double loop design where the outer loop creates new trn/val/tst data divisions.

8. If performance is still unsatisfactory, you may have to increase the size of your training subset.

Hope this helps.

Thank you for formally accepting my answer

Greg

Related Solutions

MATLAB: Layer in neural network

Two of the most important uses for Neural Networks are regression(function approximation, curvefitting) and classification/pattern-recognition.

The best way to design NNs is via the Neural Network Toolbox

In general, the best function to use for regression is FITNET. Start with the help and doc documentation using the command line commands

 help fitnet
 doc fitnet

Next check out some of my posts on the NEWSGROUP and ANSWERS. Search using

 greg fitnet

In general, the best function to use for classification/pattern-recognition is PATTERNNET. Start with the help and doc documentation using the command line commands

 help patternnet
 doc patternnet

Next check out some of my posts on the NEWSGROUP and ANSWERS. Search using

 greg patternnet

Finally, use MATLAB data to practice with

 help nndatasets
 doc nndatasets

If you have problems, it is easier for us to help if you use those datasets.

The help and doc examples use as many default settings as possible. Therefore my best advice for beginners is to begin your designs similarly with one exception. Since by default, the trn/val/tst data division and initial weights are chosen randomly, it is important to know the initial state of the random number generator. The best way to do that is for you to set it to your favorite state before using TRAIN. If designing multiple designs in a loop, initialize RNG before the loop.

 help rng
 doc rng

The default topology is I-H-O ( H=10 default) for I dimensional input vectors and corresponding O-dimensional Output target vectors. The one hidden layer contains H=10 hidden nodes containing tansig (tanh) transfer functions.

The one hidden layer net is a universal approximator. Increasing the number of hidden nodes increases the ability to model more complex functions. However using more than needed decreases the ability to operate well on unseen data. The only reason to use more layers is if you want to decrease the total number of hidden nodes or, to try to model subclasses for classification.

 hiddenLayerSize=200

Replaces the 10 tansig hidden units with 200. Typically, this is a ridiculous number.

 net.divideParam.trainRatio=70/100 
 net.divideParam.valRatio=15/100 
 net.divideParam.testRatio=15/100

Divides the data into training, validation and test subsets with the ratio 70/15/15. However, this is the default ratio. Therefore those statements can be deleted.

Hope this helps.

Thank you for formally accepting my answer

Greg

MATLAB: How to specify the transfer function for the hidden and output layer

Use fitnet for regression/curve-fitting or patternnet for classification/pattern-recognition. Both call feedforwardnet.

2 layers are sufficient: 1 hidden + 1 output

Standardize(0-mean/unit-variance via zscore or mapstd) inputs and regression outputs.

Remove or modify outliers

Convert classification targets to 0-1 unit vectors via ind2vec

Accept the fitnet default transfer functions tansig/purelin

For classification the default tansig/softmax transfers are the better choice

The best way to choose initial weights and number of hidden nodes is via trial and error.

I have posted many double for loop examples. Search the NEWSGROUP and ANSWERS

 greg fitnet  Ntrials    % for regression
 greg patternnet Ntrials % for classification

Hope this helps.

Thank you for formally accepting my answer

Greg

Best Answer

Related Solutions

MATLAB: Layer in neural network

MATLAB: How to specify the transfer function for the hidden and output layer

Related Question