Marius Myburg's Weblog

Farve System Modelling Studio is at Revision 51, and it is looking good. Everything is coming together nicely; the IDE is starting to look good, my scripting language FarveScript is great, I am starting to build up a collection of generic Specialization modules including ones for math, graph plotting, general signalling, and the Extensive Connectivity specialization is finally becoming more mature (that is a complete understatement – Farve EXCON kicks any neural net’s butt even now). Yes, I am happy because I have really worked very hard, for very long on Farve and it is starting to pay off. For me, revision 51 marks a kind of solid stepping stone, from where my whole vision can be made to manifest some day.

Oh, and I now have 8GB of RAM for my PC – still not enough to hold a moderately complex EXCON net, but it’s a start.

I am planning to release an Alpha version of my System Modelling Studio software in June, so look out for that. I can not promise that I will make that deadline though, but that is the plan currently.

Yesterday, I upgraded my PC to have 4GB of RAM, because my AI system (Farve) was dying when instantiating large cubes. Well, with 4GB RAM it gets much further – but I need more man, I need MORE!!!! The human brain has millions of neurons and billions of synapses. I need MORE!!!!

You think the sun is hot? It’s NOTHING.

http://blogs.discovermagazine.com/badastronomy/2009/03/24/the-hulking-sky/

A colleague (Eduard – now you are famous too) sent me a link to an article about the limitations of current AI, and what AI is to become. It is a very simple but  good explanation of the problem that Farve is designed to solve:

http://news.softpedia.com/news/Conceptual-Learning-Is-the-Future-of-A-I-107186.shtml

Conceptual Learning Is the Future of A.I.

Robots must take non-verbal commands as well

By Tudor Vieru, Science Editor

19th of March 2009, 08:55 GMT

Despite the fact that, over the years, formal logic has brought robotics a long way to the point where it is now, it may be that this approach to making smarter robots might no longer be sufficient anymore. That is to say, it works fine for beating someone at chess, or for matching web pages to search queries, but it has no use in making robots that are able to associate concepts with real-world objects. It takes a lot more than extensive programming to drive a machine to go fetch a beer from the fridge, researchers say.

This example is one of the best there are, for a simple reason, namely that it combines a lot of actions and a great deal of associations. That is to say, when you ask a robot to go get you a beer, it first needs to realize what a beer is and where it would usually be located. Then it would have to get to the fridge and figure out that it has to open the door before it can help itself to a can. It would consequently have to distinguish between soda and beer and take the latter in such a manner that it doesn’t crash the container.

When it would bring it to you, it would need to be able to deliver it in your hand, and not just simply drop it in your lap. And all of these actions it would have to do would be commanded by a simple line, such as “give me a beer.” No matter how much formal logic is involved, this level of coordination could never be achieved. It may be obtained for pre-defined actions such as this one, but if someone asks the machine to bring over a shovel, it will not be able to do so.

Future robots need to be able to learn from experience and to figure out for themselves the actions they have to perform in order to accomplish the command. “People realized at some point that you can only get so far with a logical approach. At some point these symbols have to be connected to the world,” MIT Media Lab AI researcher Matt Berlin explained. “People learn what a word means in a truly grounded way,” he told LiveScience.

“As humans, we can detect where there’s shadows, colors and objects. That has proven extremely difficult for robots,” Brown University robotics expert Chad Jenkins added. Most researchers in the field are currently trying to develop technologies that would allow the machines to see the world like we do, in all three dimensions, and to make them distinguish between indoor and outdoor environments. All of these things come natural to us, but are next to impossible to implement in them.

This weekend I implemented a new namespace into my Farve Extensive Connectivity Engine.  Behold, the killer namespace:

ComplexComputation.Farve.Lorena.ExtensiveConnectivityEngine.Systems.NULL.Complex 

The namespace ’Complex’ currently contains two class definitions; Plane, which is a 2D ’sheet’ of connected-to-nearest-neightbours neurons, and an extension of Plane, called Cube, which consists of multiple Planes stacked on top of each other, with Plane-to-Plane neuron connections.

Dynamics and training works extremely well in small cubes consisting of up to 1000 Neurons (10×10x10) and 19152 Synapses.  But my PC runs out of memory when I try to create a (100×100x100) Cube  (1 million neurons).

My computer cannot handle it, so I will have to get more RAM. In advanced AI, hardware is the bottleneck. Luckily RAM is cheap.  So I am getting 4GB. 

I will push the hardware I can afford as far as possible – but I must say that to see even a small 1000 Neuron, 19152 Synapse Cube actually work and learn is inspiring. Farve is coming together very nicely.

My ‘flagship’ Farve application, System Modelling Studio, is almost demo-ready.

I recently contacted that SIAI (http://singinst.org/) to ask them for some advice on how I should proceed with my AI system, Farve. I have done all this work, I have an artificial intelligence system that, quite frankly, is unmatched… so what now?  Where do I get money for research and development?  I asked the SIAI if I could perhaps give them a demonstration of Farve and let them decide if they want to help me with R&D.

Everything was going very nicely, I was planning a presentation for the SIAI, they wanted one of their AI researchers to contact me to chat about the more technical aspects of Farve… things were going well.

Then I remarked in an e-mail that, I will give them a presentation with some technical details, but in the presentation I will give them, I will not give away all of my ’secrets’ as it were.  I said that I would provide the SIAI with a technical explanation without divulging all of the details of more than 10 years of personal effort.

This is reasonable, right?  This is my work, my time, my effort, and my passion – I can not just give everything away to some organization on one sunny day.  But then one of the people at the SIAI I was corresponding with replied so (I quote):

Honestly, if you are trying to protect your idea what are you even talking to us for?

We’re here to make sure things develop safely, not for some pissing contest over credit.

One error here and you don’t get credit for the idea that ended life on earth, just dead like the people who didn’t have the idea.  If you are worried about trusting us with dangerous information that would be a different thing, but that didn’t sound like your concern, and in any event, in terms of responsible handling of dangerous information, if you aren’t going to trust us who ARE you going to trust?

Nice, isn’t it?  Can you believe the guys rudeness?  Moreover, can you believe that he expects me to give the SIAI my work, just like that?  And the guys choice of words clearly tells something about his personality too.

I think my response was justified:

Michael I see no reason for this rudeness. You misunderstood my e-mail, which is fine, but in my opinion you replied extremely rudely, which is not fine. This is not about a childish ‘pissing contest’, to use your words.  I have a technology that I believe is valuable, and that is why I approached the SIAI. Sorry if I wasted your time. I will look elsewhere for possible help, and in the meantime I will continue to develop my ideas on my own.

Unless I receive an apology, the Singularity Institute for Artificial Intelligence is getting nothing from me; not Farve, not a demonstration, nothing.  I choose to work with people with good manners. I don’t need to prove anything to anyone – I am simply working hard on one hell of an AI system, because that is what I love to do. It is not for recognition, it is not for money – it is simply a passion, I know AGI is inevitable, and I try to advance the field if I can.  To have help or funding for R&D would be brilliant, but it is less important than simple good manners.