Counterpropagation Network
The counterpropagation network is a hybrid network. It consists of an outstar network and a competitive filter network. It was developed in 1986 by Robert Hecht-Nielsen. It is guaranteed to find the correct weights, unlike regular back propagation networks that can become trapped in local minimums during training.
The input layer neurodes connect to each neurode in the hidden layer. The hidden layer is a Kohonen network which categorizes the pattern that was input. The output layer is an outstar array which reproduces the correct output pattern for the category.
Training is done in two stages. The hidden layer is first taught to categorize the patterns and the weights are then fixed for that layer. Then the output layer is trained. Each pattern that will be input needs a unique node in the hidden layer, which is often too large to work on real world problems.
More information:
Nice document with simple, clear explanation and source code for a counter propagation neural network
Source code example counter propagation network to determine angle of rotation
Application of a Counter Propagation Neural Network for Star Identification (nice graphics showing counter propagation network and clear explanation (pdf)
3 Responses to 'Counterpropagation Network'
Leave a Reply
You must be logged in to post a comment.
Thank you for this information, but how to combine the hidden layer (competitive layer) with the output layer (outstar structure by Grossberg)?
and for the training process (two stages?).
glanny
29 Feb 08 at 12:07 pm
So sorry this got totally lost.
I did some digging and I think you’ll find the top link has a clear example and explanation – better than I could do.
herself
1 Mar 08 at 12:12 am
Hi,
I need clear explanation and source code for counter propagation neural network that is being provided at 1st option above which i am not able to access could you please provide me it as soon ,since i am doing my M.Tech project “Multi class classification” with different strategies of neural networks
1.convolution neural network
2.Counter propagation neural network
thanking you.
aryan
17 Aug 10 at 1:05 am