[Korrelan]

@Ranch

I was wondering, is that morphed picture its eye?   I was wondering why its morphed because ur only looking at 2d things.

The model recognises 3D objects as easy as 2D, it’s the retinotopic mapping of my early attempt at equivalent visual cortex (V1).

http://fourier.eng.hmc.edu/e180/lectures/v1/node3.html

Evolution has some wicked ideas; this is why I try to stick as closely to the human schema as possible, you often get something for free.  This type of mapping negates 90% of the invariance problems. 

It’s also important… if my brain theory/ schema is going to used to replace human cognitive systems, upload human consciousnesses into the cloud… and eventually rule the world… hehe. 

In a nutshell, the human eye has about 130 (ish) million rods/ cones.  The high resolution image is encoded without loss of detail by the retinal ganglion cells and fed into the million or so nerve fibres in the optic nerve/ tract.  This passes through the optic chasm and arrives at the back of the brain (V1) where the signal is un-packed and mapped back into a one for one representation of the retinal image on the visual cortex.  Just think about that… that’s a 130:1 real time encoding compression ratio, and 1:130 decoding at the brain end with no loss of data/ resolution… I thought wow.

The morphed picture is my version of this compression/ mapping process, and how the receptive fields are mapped on V1, it was crude but it worked, and it was probably the first time I truly understood the brain… well my version anyway lol.

 

[ranch vermin]

Impressive.

Funny you said about the eye compressing the data from the eye.
I wonder how lensed together you can get photons and then delens them apart, for an optical computer.

[LOCKSUIT]

How does an eyeball compress/encode/map?
Nets do that...
Do ganglion cells do that? There a net at back of eye? It's does a diff type of encoding? What sort? There's only 1 i know of that I'd use.

And why if the encodin net doin it thing is in the brain and only decodes it once arrives? Maybe it uses it before getting to V1 but then that's sorta same thing...

[Korrelan]

Do ganglion cells do that? There a net at back of eye?

Yes lock, your retinas are a type light detecting neural net. The ganglion cells are the output neurons.  The retinal ganglion cells compact and change-encode the image falling on to you retinas and send it down the optic nerves, this then connects to the Geniculate body. At this point the left/ right tracts from each eye are combined, so the images from the left halves of both eyes are sent to one half of the visual cortex.

https://en.wikipedia.org/wiki/Optic_tract

And why if the encodin net doin it thing is in the brain and only decodes it once arrives?

It’s for physical reasons; your eye has to be able to move, so evolution created/ connected a thin cable from the back of your eye leading to the Geniculate body.  The bandwidth of the cable is smaller than the retinal data stream, hence the compression required.

 

[LOCKSUIT]

Can you tell us all of the wicked techniques the biological human brain uses?

I know one straight off the bat; hierarchy

Do you know any more?

[LOCKSUIT]

Evolution invents wicked ideas because it had so long and because it is forced to do simple things that result in complex things. In fact everything is a hierarchy, brain and body. DNA/1 cell becomes a full fledged human monster because of a possibly simple mechanism on a higher level yes yes, sorta like python vs machine code, and the human brain has only a few properties like a one type of same cortex, reward chemicals that spread.
I think we will soon see organism creation using small hierarchical processes using cells, or possibly efficient and quantum simulations. I feel that anyone who tries can do it in a fast amount of time ex. 5 years. We'd either simulate or make new cell that multiples or mutate a unspecialized cell we already have (or for nano fog god creation, however probably unlikey; they'll be able to create particlistic tools using super fast nano arms/foglets like do dododo do dodo dodododo because 1)light/jet impulse speed and 2) nano arms/foglets take less time to reach destination than a human elbow takes).

[DemonRaven]

Sorry but you keep forgetting one little fact our universe is not eternal neither is our sun or earth. Computers and robots break down. Chatbots are still needing to be supervised. IF anything the Age of AI will be someone programming a computer and pretending to be the all mighty singularity. The only thing resembling real AI are the neural networks which are based on brains. But even they have not gotten conversations down pat. I can spot a chatbot within five mins of speaking with them. It took nature millions of years to create us  and we think we can do it in a few decades? Talk about being arrogant.

[ranch vermin]

Sorry but you keep forgetting one little fact our universe is not eternal neither is our sun or earth. Computers and robots break down. Chatbots are still needing to be supervised. IF anything the Age of AI will be someone programming a computer and pretending to be the all mighty singularity. The only thing resembling real AI are the neural networks which are based on brains. But even they have not gotten conversations down pat. I can spot a chatbot within five mins of speaking with them. It took nature millions of years to create us  and we think we can do it in a few decades? Talk about being arrogant.

It may be a little unethical, but I dont think its arrogant to say we can actually do it,  it should have happened decades AGO.   I think something wierds up with the technology department, in my opinion we are behind.   I watch the old computer chronicals videos on youtube, and they had an A.I. department in the 70's that just got nowhere....    it seems like a bit of malpractice in my mind.  what a pack of time wasters. 

[DemonRaven]

The boldness and naiveness of youth.  The first chatbot eliza was created in 1964-1967 . I was only 9 years old then and i am now 60 years old. That is how long they have been at this. Good people  many of them with PH.D's have been at this a lot longer then you have and they are still not there yet.

There is are three things you have to realize about language.1. Language is learned by hearing, exploring and watching or environment.  So in order to get a true understanding of what they are saying they would have to become a robot. 2. Language is constantly evolving and not static. Humans are always making new little words and altering the way they say things. That makes it hard for a chatbot or a robot to keep up and that is with one language. Now try programming all of the languages in it. Even google translate screws up and it still doesn't have all the languages. 3 chatbots/AI do not have common sense so it is easy to deceive and trick one into making a mistake and giving itself away.  Those things have to be overcome in order for it to truly be able to speak and understand us.

[LOCKSUIT]

Let's make sure not to flood korrelan's special thread too much.

I agree ranch, its long overdue. I've seen images of "pops" working on the big ol PC in like 1950 on AI just after ey made the darn thing! Talk about 1950 explosion. They said "anytime now". Some ol timers were already codin wizes nuts like teens.

They say my HDD will lose the data but I still got it from age 13 now I'm 23. All my anime photos are preserved ;p. Backup. Create superorganisms that don't die. Use redundancy. We'll escape the dying sun long before that happens. I'm already on it.

[ranch vermin]

Its not even speculation on my behalf,  the "machine learning" stuff out now outpaces the old ways of doing things,  by others,  not even including my own personal work.
What stopped the old programmers from knowing the almighty power of assignment (=)   its simple,  its not complicated,  but the olden days nothing developed much, and they couldnt make computers for nuts either for not much real decent reason.

We have gpu's these days,  but why did it take so long?  Doesnt make sense to me.

Are the kids these days an evolution of the old mind?    How much of this "evolved dna" is given off to africans for them to feel unrighteously superior?   

[Korrelan]

Ooo… lots of posts…

IF anything the Age of AI will be someone programming a computer and pretending to be the all mighty singularity.

I hope the intelligence of the AGI will speak for its self.  As you say you can recognise a chatbot with five mins of conversation, surely people could recognise a true intelligence. 

One definition of the Singularity is machines designing ever more intelligent machines, it would be pretty hard to pretend/ fake this type of scenario.

It took nature millions of years to create us  and we think we can do it in a few decades? Talk about being arrogant.

As this is on my project thread I can only assume it refers to people like me.

True story… Two farmers stood in a field next to a beach in 1903 watching two blokes with a weird looking contraption.  One says to the other… ‘Just look at Orville and Wilbur Wright the idiots, it took millions of years for birds to be able to fly, what arrogance.’ Though I suppose they wouldn’t know about evolution and probably attributed it to some deity, which only amplifies my point.

My point being that ‘arrogance’ can be an offensive label, now were the Wright brothers wrong for knowing how to solve the problem space, or was it the ill informed farmers?

Luckily what seems impossible to one person/ group is possible to another… the human race wouldn’t have got this far if this was a fallacy.

Those things have to be overcome in order for it to truly be able to speak and understand us.

I agree, and that’s the whole point of my project, please take the time to read it if you haven’t done so.

I think something wierds up with the technology department, in my opinion we are behind.

Nothing as changed… except now we have the internet.  In the past groups of scientists/ researchers would work within the closed sphere of their facilities.  They had their own reference books/ libraries and fellow associates and that was it. It would often be a big deal to have a visiting professor, or an expert visit and give their opinions and ideas.  This is where the whole concept of white papers, conferences and lectures arose from.  But now we have the internet which supplies up to date information, the latest research, etc. 

Like a human brain it’s the interconnectivity that’s making all the difference, for the first time in human history we are not isolated, anyone can access the ideas/ research of anyone else.

Are the kids these days an evolution of the old mind?

I think that’s a totally different problem and again it involves the internet. The internet is a double edged sword, whilst it is allowing science research to flourish; the interconnectivity is also allowing the social aspects of society to… I’ll stop there… this is an AI discussion after all.

 

[LOCKSUIT]

Universe may end, but I'll try living for as long as I can. Just like you, you want to stay alive still DemonRaven even though you think you can't forever.

I see everything now like this: Things are possible if done right. Look around. You *can make a dishwaser, or make AI, or escape Earth in next 5 years if done *exactly correct. If you never invented much, all you know is "that's impossible/rare".

[LOCKSUIT]

Korrelan, an AGI has to think on its own and have desires, so what do you call neurons that fire on their own? In my work, I call them self-ignition, sometimes I call them self-activations. It's like those thoughts in the morning that pop up "ah, my AI work...back to the computer!, (after some coffee)".

[Korrelan]

so what do you call neurons that fire on their own?

I think the brain is a deterministic system, nothing is truly random and it only seems random at the level of abstraction / of the observer.  If you drill down deep enough you will discover the underlying mechanisms.  So Neurons in the human brain never actually fire on their own, it may not be obvious to the observer/ researcher but they are firing to a stimulus (or lack of) of some kind. 

The environment a neuron resides in does seem extremely noisy, if you single out a neuron or group without understanding the context of their receptive fields and communication protocols.

Time for an analogy…  If you stand at a busy intersection/ junction on any road, looking at the passing traffic it seems very busy/ data noisy, can you figure out what each person is doing or where they are going?... they must be doing something, going some where, can you figure out the local economy from this single small data point.

Neuroscientists have the same problem, that’s why they are striving to monitor larger groups of neurons, the problem they are going to hit however is that some of those cars are travelling to local destinations, some however are just passing through on a journey of hundreds of miles.

My point being that just because a Neuron seems to fire sporadically/ randomly, it doesn’t mean it is not responding to a stimulus.

It's like those thoughts in the morning that pop up

The process is not random.  A thought is a culmination/ compounding of many smaller facets. In other words things only seem to pop into your head because you are not consciously aware of the underlying, sub conscious mechanisms driving/ guiding your consciousness.

Everything is influencing your decisions, the time, the temperature, your body, experience, previous repetition (episodic memories), prediction, etc so many variables.  If you were to get out of bed and stand on a Lego… the thought of working on your AI would be the last thing to pop into your head.

This is another way a brain is very different to a computer.  A computer has functions, sub routines, single algorithms for doing single jobs.  With the brain however each function is a blending derived from the summation of hundreds of smaller similar functions doing a similar job.

So a single finger joint muscle does not just have one network governing it, it might have ten, all slightly different taking information from different sources, each driving sets of fibres within the same muscle.  Ever get the shakes after a night on the booze?  Basically some of the many networks that control your hands are out of sync… still drunk… hence the course movements/ shakes.

Every so often I like to remind readers that any information I impart is based on my current understanding of main stream academic research, and is often adjusted/ updated with both personal insights and findings through my research.  A** coverage lol.