First Neural Network Using XOR

I am trying to implement a simple XOR network. All is okay once the input and target data has been setup, but as soon as I try and train the network I get the Neural Network Training Tool window open, but the "stop training" and "cancel" button are shaded out with "minimum gradient reached". As soon as I try and simulate the network, the XOR_NET_output data is wrong and there seems to be error data within the XOR_NET_errors.

I can provide more data if necessary.

[Shashank Prasanna]

Since this is a fairly simple setup, could you share your data and the lines of code you've written? It will be easier to look into the issue.

Answer 1

[Greg Heath]

I have many posts on the NEWSGROUP, ANSWERS and comp.ai.neural-nets re XOR. Most can be retreived by searching on

greg xor

The minimal configuration has a 2-2-1 topology with Nw = (2+1)*2+(2+1)*1 = 9 unknown weights to be estimated with only 4 equations. Consequently, there are an infinite number of solutions.

Nevertheless, I recall a success rate of only ~ 70% when training from a random set of initial weights generated by MATLAB's default NW algorithm.

So, just try 10 or more different random weight initializations. You should get at least 5 successful solutions.

Hope this helps.

Thank you for formally accepting my answer.

Greg

Answer 2

[Mohan]

The implementation of the XOR with neural networks is clearly explained with Matlab code in "Introduction to Neural Networks Using Matlab 6.0 " by S. N. Sivanandam, S. N Deepa