Neural Nets with AI::NNEasy

gri6507 has asked for the wisdom of the Perl Monks concerning the following question:

Fellow Monks,

First off, thanks for all of your responses to my earlier node Neural Nets. Based on your suggestions and the type of work I would like to do, i decided to go with AI::NNEasy. As the name suggests, the interface to the module is truly easy and seems to be rather fast.

I do, however, have a question. In an attempt to do a hybrid AI::NNEasy and Matlab solution, I would like to use AI::NNEasy to train the net and get the weights and biases, then start my Matlab net with those values. Is there a reliable hack (apparently there are no built-in methods) to extract the weights and biases?

Thanks again.

Comment on Neural Nets with AI::NNEasy

Replies are listed 'Best First'.
Re: Neural Nets with AI::NNEasy by g0n (Priest) on Apr 11, 2005 at 14:23 UTC
Not that I'm aware of. gmpassos may have developed something, but I don't think he uses MATLAB for neural nets. AI::NNEasy uses Storable to save the network object, so you should be able to write a script to read the object in from the file, identify the weight values and write them out in the format you need. Update: Data::Dumper is likely to be useful doing this. g0n, backpropagated monk	[reply]