f



Neural Network Architecture with MATLAB Neural Network Toolbox

Dear neural network experts,
I would like to make a neural network model for a situation
when some of the input variables can interact, while
others don't have any interactions. This information is
based on logical considerations and I have to include it
into the architecture of the net.
For example, I have 3 different groups of input variables.
Within each groups the variables are disjunct (i.e. no
interaction). Variables of different groups may interact.
This would result in either
- restrictions to some weights of a network consisting of
completely inter-linked neurons (i.e. some weights have
to be zero), or
- a combination of neural networks, one for each group of
variables and one combining the groups (nested nets?).

This network needs to be created, trained and simulated.
How can I implement this based on commands of the neural network
toolbox (newff?, trainlm?)?

I would be very happy if you can point me to an example or
similar that demonstrates how the commands have to be used
in this situation. 
 
Many thanks in advance! 
Steffen

0
9/19/2005 5:35:12 PM
comp.ai.neural-nets 5773 articles. 2 followers. tomhoo (63) is leader. Post Follow

1 Replies
1396 Views

Similar Articles

[PageSpeed] 16

mathneural@web.de wrote:
> Dear neural network experts,
> I would like to make a neural network model for a situation
> when some of the input variables can interact, while
> others don't have any interactions. This information is
> based on logical considerations and I have to include it
> into the architecture of the net.

THe MLP and RBF are universal approximators. Interaction
information is automatically taken into account by the
nonlinear transfer functions. Why do you think you "have to"
include interaction information in the architecture?

> For example, I have 3 different groups of input variables.
> Within each groups the variables are disjunct (i.e. no
> interaction). Variables of different groups may interact.

Strange. It make more sense to have within-group interactions
but no between-group interactions.

Two quick and dirty ways to identify significant interactions
are

1. a. Estimate the correlation coefficient matrix of the
      variables and their cross products
   b. Estimate the corresponding confidence intervals
   c. If a correlation coefficient is smaller than half
      of its confidence interval, set it to zero.
2.Use stepwise linear regression on the inputs and their
  cross-products.

> This would result in either
> - restrictions to some weights of a network consisting of
> completely inter-linked neurons (i.e. some weights have
> to be zero), or
> - a combination of neural networks, one for each group of
> variables and one combining the groups (nested nets?).

This could be done with different layers of a single
NN. However, I don't think the Toolbox algorithms allow
learning with weights constrained to be zero. Therefore you
would have to write your own algorithm. Cosequently, combining
networks would be easier.

> This network needs to be created, trained and simulated.
> How can I implement this based on commands of the neural network
> toolbox (newff?, trainlm?)?
>
> I would be very happy if you can point me to an example or
> similar that demonstrates how the commands have to be used
> in this situation.

I don't believe the zero weight constraint feature is available.

It might help if you first learned as usual and then zeroed out
insignificant weights.

Hope this helps.

Greg

0
Greg
9/20/2005 12:12:20 AM
Reply:

Similar Artilces:

neural network predictiver controller in neural network toolbox
I got some question that would like to ask about the minimizaton routine in neural network predictive controller in neural network toolbox. 1. Is there any method to calculate the control signal which minimize the performance index beside using Simulink? Any Matlab command which is relevant for minimization of performance index? 2. What is the teory for minimization routine like csrchbac? Is there any reference that I can refer? I am looking forward for any reply from any other Matlab user. Thank you very much for spending your time to read my message. Thanks again. Hi, G.H.Heng you work ...

neural network: Recurrent Neural Networks
Hi Greg and Anders, thanks for your help of my backpropagation topic. Now I switch to recurrent neural networks. My question: 1. Does it the recurrent neural networks need to divide input data into training, validation and testing 3 sets? I tried to convert my previous brackpro code into recurrent but it prompt out the error that the "train" code section has too many input arguements, which include VV and TV... Felix Hi I tried to use Elam network. One strange result came to me. I used 10 hidden units it shown that the preformance goal is not met. and the mse is about 8.5 . Then ...

How can I create a neural network for data classification using the Neural Network Toolbox?
I want to create and train a neural network which can classify the input data. In my case, I would like to classify the voice/speech input signal. For your information, i had already done the feature extraction stage for the signal. Now, i have a problem how to recognize/classify using neural network.Please help me ..thank you. "fikrulhakim" wrote in message <htdtq1$irq$1@fred.mathworks.com>... > I want to create and train a neural network which can classify the input data. In my case, I would like to classify the voice/speech input signal. For your information, i had alr...

Probabilistic Neural Network and Radial basis function Neural Network
I am a final year student in B.Tech Information Technology. I am doing a project related to object recoginition using Probabilistic Neural Network and Radial basis function Neural Network. Please send me the running code of these networks. Thanx ...

Request : Introductory Material on Neural Networks and Fuzzy Neural Networks
Hey All, I am interested in Neural Networks and Fuzzy Neural Networks. My school library isn't staffed with the books that were mentioned in the FAQ. Other than the wikipedia article, could anyone tell me where I could get good online introductory material about Neural Networks and Applications of Fuzzy Logic in Neural Networks. Thanks a lot, Devdatta http://www.ai-junkie.com has some pretty good info :) "dev" <dev.akhawe@gmail.com> wrote in message news:1163341237.990631.232940@h54g2000cwb.googlegroups.com... > Hey All, > I am interested in Neural Networks and...

Neural Activity in Neural Network
Hi, Sorry for my "english"... I trained a neural network and i want to study it. This network have one hidden layer and I want the activity in tis hidden layer... but i don't make that... I try that : %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% load net_RDN input = ones(1,19); % pour voir la repr?sentation du r?seau %view(net) % mise en forme du vecteur d'entr? (en ligne) if size(input,2) == 1 input = input'; end % calule de l'activit? des neurones de la couche cach?e num_layer = 1; % indice couche cach?e for a=1:net.Layers{num_layer}.size activity.HL(a) = f...

From Matlab neural network toolbox to matlab code
Hi All, I have been working with neural network toolbox in matlab 7 for the calibration and validation optimization problems. but i wanna to do it with matlab code. is it possible to transform the things that happen in toolbox, training process and adapting outputs get via code.. is there anyone who transform the things happen in toolbox to codes. "murat " <muratt59@gmail.com> wrote in message <iabd8c$8pc$1@fred.mathworks.com>... > Hi All, > > I have been working with neural network toolbox in matlab 7 for the calibration and validation optimization...

Matlab neural network toolbox
Hello, Anyone out there have personal experience with the neural net module for Matlab? I'd appreciate hearing about your experience, since I'm considering the purchase of Matlab and the NN package. On 29 Jan 2004 10:06:38 -0800, cook_jim@yahoo.com (J Cook) wrote: >Hello, > >Anyone out there have personal experience with the neural net module >for Matlab? I'd appreciate hearing about your experience, since I'm >considering the purchase of Matlab and the NN package. I've just starting using it and was able to get up and running very quickly. There&#...

Som toolbox (neurol network)(neural network clustering)
hi eveyone &#305; wanna clsuter iris data set with som but in matlab &#305; cant do it or &#305; cant understand how it is making pls somebody help me sincerely ali On Apr 8, 8:59=A0am, "ali " <rebelfi...@gmail.com> wrote: > hi eveyone > > &#305; wanna clsuter iris data set with som but in matlab &#305; cant do = it or &#305; cant understand how it is making pls somebody help me > sincerely > ali 1. Use a different font. There are strange characters in your post. 2. Post relevant code and error messages. Hope that help...

Calling Neural Network Toolbox from .Net
Hi, I wonder if a pre-trained neural network object can be transformed into a dll using matlab compiler(MCC). I want to simulate (sim(net,Inputs)) a multilayer perceptron network object from .Net. But i wonder how to export this object to a dll. Any help ! Thanks Uzair Ahmad I found it easy to write my own neural network object in dotnet and then bring in the fitted parameters. On Mon, 5 Jun 2006 03:56:01 -0400, "Uzair Ahmad" <uzair@oslab.khu.ac.kr> wrote: >Hi, > >I wonder if a pre-trained neural network object can be transformed >into a dll using matlab com...

matlab neural network toolbox #5
hello; matlab neural network toolbox has NN predictive controller (system model+controller)and model reference controller blocks(system model+controller) does any one know how can i use these bocks only as a system model block because i did not need for neural controller thanks ...

NN(Neural Network) Toolbox MATLAB
Hello all, I am new to the forum and this is my first post. I am undertaking a final year project which involves the use of MATLAB. In particular I need to use the NNToolbox. I am having great difficulty using it and would like some advice from more experienced users to help me. Ok here is a brief description of my assignment: 1.I have transcribed a list of names into phonetic format using the IPA(International Phonetic Alphabet). Each name is made up of phonetic symbols. In total there are 54. 2.Each symbol is numbered: the first as 0 and the last as 54, working like an index. 3...

matlab neural network toolbox #3
hello; matlab neural network toolbox has NN predictive controller (system model+controller)and model reference controller blocks(system model+controller) does any one know how can i use these bocks only as a system model block because i did not need for neural controller thanks ...

matlab neural network toolbox #4
hello; matlab neural network toolbox has NN predictive controller (system model+controller)and model reference controller blocks(system model+controller) does any one know how can i use these bocks only as a system model block because i did not need for neural controller thanks ...

Web resources about - Neural Network Architecture with MATLAB Neural Network Toolbox - comp.ai.neural-nets

Coding the Architecture
Coding the Architecture

International style (architecture) - Wikipedia, the free encyclopedia
The International style is a major architectural style that emerged in the 1920s and 1930s, the formative decades of Modern architecture . The ...

Come to Flemington for great food, including an American barbecue festa, its architecture and the 'other' ...
The place famous for its racecourse and showgrounds also boasts a great river and a new festival.

Cox Architecture qualifies for Tokyo Olympics in winning swimming pool bid
Cox Architecture will consult on the Tokyo 2020 Olympic aquatic centre, three years after losing the main race.

AMD Reveals Polaris GPU Architecture: 4th Gen GCN to Arrive In Mid-2016
... for display technologies – DisplayPort, HDMI, Freesync, and HDR – and how the company would be laying the necessary groundwork in future architectures ...

AMD’s new graphics architecture is called Polaris
The rumours were true: AMD's new graphics architecture is called Polaris (it previously went under the codename Arctic Islands), it's based on ...

Rolls-Royce gets down to testing new aluminum architecture
Filed under: Spy Photos , Rolls-Royce , Luxury Rolls-Royce is testing a new aluminum architecture that's set to underpin not only its upcoming ...

Brand Architecture’s Impact On Brand Value
Brand architecture often comes down to an evaluation of tradeoffs. In my experience, there’s rarely a cost-free benefit or a no-foul cost. That’s ...

AIA: "Architecture Billings Index Ends Year on Positive Note"
Note: This index is a leading indicator primarily for new Commercial Real Estate (CRE) investment. From the AIA: Architecture Billings Index ...

​Chicago launches its first Architecture Biennial
The city celebrates both old and new, with the latest designs of leading architects redefining space, function and community

Resources last updated: 2/7/2016 7:28:05 PM