[Art]

I didn't think the robot was "stuck" toward the end but rather that it had "learned" not to collide with any of the walls. This is what I took to be the initial premise of the experiment, correct me if I'm wrong.

Are the prox sensors adjustable by any onboard means? If might make the test more interesting if it could come a bit closer toward the walls before reversing direction.

I really enjoyed your video and thanks for sharing!

PS - As a side note, there is a Home Made Robots section under the Topic heading of Robots. This and other such related videos might gather more attention and feedback if placed there.

[Korrelan]

By the way korrelan, what do you mean by Human Neural Architecture?

I’m basically back engineering the human brain.  My design incorporates neurons , glial cells, basket cells, astrocytes, axons, pre/ post delays, myalin, synapses, 12 neurotransmitters (GABA etc), 6 six layer cerebral cortex/ lobes, cortical columns, mini columns, hippocampus, cerebellum and most importantly time, plus many more virtual equivalent brain structures.

I first wrote the 3D CAD software where I could easily design and tweak my ideas; this now incorporates an AI for voice and gesture control.  The whole system is massively parallel and designed to run on multiple processors/ machines; visual, audio and tactile sensory streams are provided by 1GB TCP pipes from separate machines; this makes it easy to expand. (Beowulf cluster)

I grow each AGI I test from a neural seed that expands in virtual space to accommodate the spatiotemporal /holographic memory that develops from maturation.  The simulation is then put through a rigorous accelerated sleep/ learning cycle (I can’t wait 5 years for each to develop) and I test the connectome.

This is a neural seed…



There are some more videos on my Youtube showing object recognition etc.

This is the visual orientation map from a Swindale monkey.



And this is the orientation map from my AGI developing in real time form visual input.
(Different orientation marker colours but you get the idea)



It keeps me occupied and out of trouble 

Just in case you were wondering I don’t delete a partially developed AGI, it just doesn’t seem right somehow.  They are saved to disk into a directory called… cryo_storage lol. 

[Ultron]

Cryo_storage nice one! 


Anyhow, amazing project! I usually see a lot of 'human neural brain architecture' or whatever projects which barely show an example or any real concept of functionality.


From a personal perspective, I believe it is much easier (relatively, must note) to mimic the human 'hardware' rather than the software. Many have done an excellent job at re-creating an artificial version of a human (or rather, humanoid) brain or a complete neural network of a worm and have shown that they indeed function as intended, however we still do not see androids walking around - why?


Maybe it is because I am not a rich or funded lad that I chose to focus my research in the philosophical area of the A.I. field, but in the end I suppose I am glad I did because not many are really approaching this problem of re-creating the brain like that - most either focus on the hardware aspect (computer clusters, deep neural networks / learning etc.) or take the two problems as one. That doesn't work, at least in my opinion.


It also makes a lot of sense - you are reverse-engineering one of the most genius 'machines' nature has ever created, and it took 'nature' quite a few millions of years to do so (please do not forget our not-so-humanoid ancestors). To make an analogy with a simple machine - if you were to reverse-engineer an old 70's computer and find out what it was used for, first you would need to figure out the hardware, and then how the hardware was used (software) - otherwise, you are stuck with a bundle of guesses on how the hardware could have been used.


To make it clear, this was written in respect to all research done in the field of artificial intelligence, but what bugs me is that most people tend to keep ideas to themselves or their own companies and aim high for profit, rather then knowledge. If this wasn't true, I would have been making BBQ on Ganimede right now.

[Korrelan]

Pleased you like my project.

Maybe it is because I am not a rich or funded

Do what I did… Start your own companies to fund your project.

first you would need to figure out the hardware, and then how the hardware was used (software)

The beauty of my approach is that the emergent connectome defines both the structural hardware and the software.  I think the brain is more like a clockwork watch or a Babbage engine than a modern computer.  The design of a cog defines its functionality.  Data is not passed around within a watch, there is no software; but complex calculations are still achieved.  Each module does a specific job, and only when working as a whole can correct function be realised.

As you can see from the examples I tested the basic theory first, proving the connectome that arises from my algorithms can understand and parse language, recognise visual objects/ scenes, pay attention, reason, predict, dream, exercise free will etc.  I am now at the stage of combining all the working modules into a whole intelligence.

I’m sure there are many routes to achieving an AGI, all routes must be explored.

[Ultron]

I like how you gave all the aspects of the system some thinking, korrelan!


On the topic of... let's call it 'default' software (this refers to your babbage engine)... I can agree that those ideas are sound and may seem somewhat practical, but keep in mind that humans (and most likely other animals with a more developed neural system) have some default or instinctive 'functions' 'pre-coded' into their memory. This might be compared with the muscle memory we are born with, such as the baby's instinct to fall on their buts (a smart choice and way to fall), which is lost as we grow up.


This might mean that there might indeed be some sort of genetic program or a biological controller hub / network of cells that learn from DNA-encoded information how to make the system work or rather, control it. You cannot deny this because there are some things in life you are born knowing 'how-to', such as eating (or rather, the part where you chew and swallow) - you don't learn that in school, do you? Also you know that crying will make your parents or keeper respond and assist you...


Let's not go in deeper... As usual, I will assume that by now you got the point?

[ranch vermin]

Wow!  I didnt respect you enough Korrelan.  (smack myself)   That really is impressive,  and looks like this other natural based system.