Here is another site with simple source code for neural networks:

http://www.ip-atlas.com/pub/nap/nn-src/

You can use Encog Machine Learning Framework available at:

http://www.heatonresearch.com/encog

You can find Java, C#, C/C++ code there

Alternatively, at http://www.codeproject.com/Articles/175777/Financial-predictor-via-neural-network you can find a project wtih source code for financial time series prediction using neural networks

Another good project with source code available for free:

http://sourceforge.net/projects/neuroph/?source=directory

WEKA is a good option if you code in Java: http://www.cs.waikato.ac.nz/ml/weka/

you can find plenty of tutorials online on how to use neural networks with WEKA, for example:

https://blog.itu.dk/SPVC-E2010/files/2010/11/wekatutorial.pdf

Thanks a lot for your responses.

@Bogdan: Thank a lot for your help.

But recently i have studies encog it offers great library for programming but it is less explanatory as book goes for complex it directly provides libraries for getting work done so i am searching for something that will describe in depth with step by step approach. Thanks again.

Matlab help is the best source also you can refer Mathworks website for the same.

Here is another site with simple source code for neural networks:

http://www.ip-atlas.com/pub/nap/nn-src/

Beginner coders should consider python. A good package to look at is neurolab:

http://code.google.com/p/neurolab/

Actually i am familiar with java and .net. i have tried encog with java but i want step by step execution so that i can look into detail how algorithm or network in working.

I have used two very good toolboxes for prediction and control, here are the links:

www.iau.dtu.dk/research/control/nnctrl.html

www.iau.dtu.dk/research/control/nnsysid.html

Normally best way will be to read a book then implements the neural network code that comes with it, but if you are in a hurry, try this one (in C#, java version also available at the https://code.google.com/p/encog-java/)

http://www.codeproject.com/Articles/16447/Neural-Networks-on-C

I would use MATLAB and MATLAB help, it provides comprehensive information on ANNs.

Choose Matlab version 2012a & above. There are examples available in the help menu for your requirement.

If you want to do more than a simple ANN I recomend MATLAB. By the way, you have neural networks libraries in different programming languages.

You can use Encog framework.

http://www.heatonresearch.com/encog

ENCOG doenst work

i have tried encog and it provides NN model and also worked for me.

It is good if you are new to NN you can design simple NN with very small coding.

Im implementing NN with Encog in C#. I have a problem with my dataset. My training example consist of four features, numbers between 50-200. I dont know if i have to normalize this dataset to values between 0-1. If yes, how? My next issue is in testing: I load one testing example (one row of four features) into double array and I put it in command: IMLData output = network.Compute(test), where test is that double array[ ]. This is not working, because it expects there IMLData output = network.Compute(pair.Input)

I have used encog with java but i don't think it will make any difference.

It is better to have normalized values you can use simple normalization technique like min-max normalization.

Now after training your network you have to provide input data and output data as MLData and then compute output for provided input

e.g.

MLData ip=new BasicMLData(array_of_inputs);

MLData out=your_network.compute(ip);

then you can simply print and see your output.

I have used encog a long ago so i exactly don't remember but thats what i found in my programs.

Thank you.

I used the guidelines and samples of Matlab. The code normally consists of three blocks:

1- constructing the ANN using Matlab command such as (newfftd) after you define all parameters such as the number of layers and neurons per layer.

2- training the ANN using command such as (train) after setting all training parameters. May be it is a good idea to perform some data messaging/scaling prior to training as well.

3- Validation which is pretty crucial in designing reliable ANN for prediction/estimation where one would use "validation data set". that differs from the "training data set". This is important.

Matlab and Encog are tools for creating NN, but what if I want to code neural network step by step without any toolboxes for example in C#. Is there any step by step tutorial?

I never tried coding in C or C++ though there are so many books on the subject of ANN in C and C++. For example:

http://www.amazon.com/Neural-Networks-Fuzzy-Logic-Book/dp/155828298X/ref=sr_1_2?ie=UTF8&qid=1396522382&sr=8-2&keywords=neural+networks+in+C%2B%2B

Hi,

you can get the NNSYSID toolbox that works with Matlab and has been created by Magnus Nørgaard.

http://www.iau.dtu.dk/research/control/nnsysid.html

Regards

Noureddine

Aside from the fabulous Encog, you can also try Sharky Neural Network (which has a realtime visual feedback to see the decision frontier).

You can also try the various NetLogo models of "Artificial Neural Networks" (generally with a visualization that could help you understand what's happening) like this one:

http://ccl.northwestern.edu/netlogo/models/ArtificialNeuralNet-Multilayer

And if you would like to get a step-by-step tutorial on coding your very own artificial neural network, there is the Coursera Machine Learning class by Andrew Ng (https://www.coursera.org/course/ml) which contains 2 weeks dedicated to neural networks. I have learned to code my first ANN this way, so I can advise this course to anyone, this is a very good introduction (the main missing stuff is the PAC learning theory, but otherwise you will learn a lot about machine learning).

Here are a few good links if you want to further extend your knowledge about artificial neural networks (but you should first know a bit of the theory beforehand! as these links will give you very useful hints on how to tweak your ANN to work better in practical tasks):

http://pages.cs.wisc.edu/~bolo/shipyard/neural/local.html

http://www.iro.umontreal.ca/~bengioy/ift6266/H12/html.old/mlp_en.html

http://www.heatonresearch.com/content/non-mathematical-introduction-using-neural-networks

If you know R you can use "neuralnet" package.

Thanks Stephan for the info

Hi,

you may find that some reference implementations with documentation could help:

WEKA: http://www.cs.waikato.ac.nz/ml/weka/ (java)

OpenCV: http://docs.opencv.org/index.html (c++)

Cheers

very good is also using of convolutional neural networks, link: http://www.mathworks.com/matlabcentral/fileexchange/24291-cnn-convolutional-neural-network-class

Few minutes ago, I answered similar question on the subject. In addition to what I mentioned there, please have a look at the ANN Toolbox as it contains samples and demos..

My answer there is:

I have used Feed-forward neural network (which is called multilayer MLP) in designing ANN-based Fault Detection & Isolation (FDI) system and for that purpose I used real data sets and the result was pretty good. I have used one data set for training purposes and another one for verification and validation ad the results match with the a third data set. Mathematical details on the quality of MLP can be found in many references and the best I have consulted is Neural Networks: A Comprehensive Foundation

http://www.amazon.com/Neural-Networks-Comprehensive-Simon-Haykin/dp/0023527617/ref=sr_1_9?s=books&ie=UTF8&qid=1423770248&sr=1-9&keywords=neural+networks

A practical book on the design is Neural Network Design (2nd Edition) by Martin Hagan et al and can be downloaded along with lots of support material from:

http://hagan.okstate.edu/nnd.html

or else you can buy hard-copy :

http://www.amazon.com/Neural-Network-Design-Martin-Hagan/dp/0971732116/ref=sr_1_2?s=books&ie=UTF8&qid=1423770248&sr=1-2&keywords=neural+networks

You can find the similar problem and the discussions on it  at this link:

https://www.researchgate.net/post/Neural_Network_prediction_problem#54dd04d0d039b175558b45a2

I hope all this help.

Here is back propagation neural network code in java with XOR example

https://kunuk.wordpress.com/2010/10/11/neural-network-backpropagation-with-java/

Dear Prashant,

Use following link to make simple coding in MatLab.

http://lab.fs.uni-lj.si/lasin/wp/IMIT_files/neural/NN-examples.pdf