Jump to content


This topic is now archived and is closed to further replies.


Wrapping Amygdala

Recommended Posts

I recently was pointed to the Amygdala spiking neural network library http://amygdala.sf.net/ for some AI stuff I was doing. I decided that PHP deserved to have an extension for this great C++ library, so I started working on it. I'm using my ext_skel replacement to ease the development, and that's nice so far (adding C++ support and class parameters, and I'll release it when I'm done, of course), but I've hit quite the brick wall... you create an instance of the LayerConstants class and set the parameters for the first layer, then create it and add it to the network, then you set the parameters in the LayerConstants object for the next layer, and so on and so forth. The problem is, I have no idea how I should keep track of what LayerConstants object I should be looking at, or how to get it from userspace. Perhaps I could just make any changes to the PHP instance of the LayerConstants change an instance of the C++ object somewhere in my code, if so, is there a way to have a function called when a parameter is changed in an extension (similar to __set() in normal PHP code)?


Looking forward to any advise on the subject.



Lord Daeken M. BlackBlade

(Cody Brocious)

Share this post

Link to post
Share on other sites

A neural network C extension would be nice to see, im thinking of working on my own one soon (just ordered that book by sara goleman on extending and embedding php) - I managed to throw together this app Neural networks for handwriting recognition using php which uses a neural network to work out what letters you have drawn using the mouse.


How would you go about training your neural networks? would this be done in PHP as well? or via some external application?

Share this post

Link to post
Share on other sites


Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.