profhugodegaris

Species Dominance, Artilects, Artilect War, Cosmists, Terrans, Gigadeath, Essays, Media, etc

Artificial Brain Software and User Guide

Artificial Brain Software and User Guide

This tab contains the source code and user guide of the work I was doing from 2008 to 2010 in China, evolving neural net modules and then connecting them up to make a network of (evolved) neural  networks, that I called an  artificial brain.

I give this work free to the world, since otherwise it will rot. Now that Im retired, I no longer do such work, concentrating now on pure math and math physics instead. It contains initially a user guide, then the software itself, that any bright undergrad or graduate student might like to compile and play with.

For those programmers interested in artificial brains, neural networks, genetic algorithm etc, you will need to understand the theory behind the code before you plunge into trying to understand what the code is doing.

Now that I have put up (in Sept 2015) my publications list on my site (1987-2010) you can download the relevant papers and study them.

The software consists of two main parts, labeled

a) PARCONE (Partially Connected Neural Evolution)

b) IMSI (Inter Module Signalling Interface)

The Parcone software evolves individual partially connected neural net circuits which are later connected up to make a network of (evolved) network modules. IMSI is the software that allows users (human BAs (Brain Architects)) to connect up Parcone modules and to run their signaling once the BA (brain architecture) has been specified, i.e. evolved and connected up.

The code was written during the years 2008-2010, i.e. in the last two years before my retirement. I have not continued the work since then, because retirement allowed me to return to my old love of pure math, and math physics. (I make YouTube lecture videos in these topics at masters and PhD levels, as well as making e-libraries for students at grad level to educate themselves for free.)

The code is written in “C” language (not C++). I didn’t bother using an object oriented language, because it was written by me, i.e. one person, who understood the whole thing, because I wrote it. The code has quite a bit of commenting so, hopefully, after you study the relevant papers, whose links I provide below, you should be able to deduce what the code is doing.

The code makes heavy use of pointers (to pointers, to pointers …). If you’re used to pointer free coding, e.g. using C#, you may be put off by the multilevel pointing data structures.

The programming environment I used at the time was Microsoft Visual C, Version 6.

If you do get hooked on what this code can do, namely evolve individual neural net modules and then hook them up into artificial brains and run the intermodule signalling, then by all means push ahead with it and modify it, to your heart’s desire. The plan all along was to design artificial brains, to encourage the creation of a new research field of “Brain Architecture”

Prerequisite Papers to be Studied before Studying the Code

Hugo de Garis, Keynote Paper, “PARCONE, A PARtially Connected Neural Evolutionary” Model Serving as the Basis for Building China’s First Artificial Brain”, 3rd Int. Conf. on Intelligent System and Knowledge Engineering, Xiamen, China, November 2008.  (link)  (link2)  (link3)

Hugo de Garis, “Artificial Brains: An Evolved Neural Net Module Approach”, in book “Computational Intelligence: A Compendium (Studies in Computational Intelligence), pp 797-848,  by John Fulcher and Lakhmi C. Jain (eds.), Springer, 2008  (link)

The Code Itself

You can download the code from (here) and cut and paste it into your own more modern programming environment. The code is readable in Microsoft Word so you will need MS Word on your computer.

Good luck. Enjoy. Have fun building a brain with maybe 1000s of neural net modules!! If you take this work up seriously for a masters or PhD degree, let me know (profhugodegaris@yahoo.com).

%d bloggers like this: