[Korrelan]

My Dad warned me there would be days like this. 

[LOCKSUIT]

Like what?

[madmax]

I think that picture is decomposing when you see it first time to the component that you previous learn to recognize, so you constantly learn to learn.But for the short time picture is full in your mind and that must be that sensing process that you explained these guesses and actions so signals go back to the retina where picture is still existing for short time and go back and forth for many times until signals disappear, or not only retina but whole way to the hippocampus is like some circuit that last for some short period of time.But in some people could last longer that is why they have photographic memory because in their brain picture last longer on that first sensing path where picture is still whole they learn better the components of the picture and collect the better picture of what they saw in their long term memory.

I just guess is the hypocampus but maybe is even before it  place in the brain that hold picture for some short time.But the key part is decomposing of the picture that works like you said whit similar connect to similar which triggers the reward. as you said, just neuron spikes that match each other is enough to trigger the reward which triggers the further matching.But how you establish chemical reward system in simulation of neuron synapses i dont know.I guess you mast have some upper system for detecting spikes that now blow some new energy for firing new synapses.But i even dont know how work these artificial neural nets. 

[LOCKSUIT]

VideoSense of walking all the way through a beautiful lengthy laboratory and thinking about the whole last 5 minutes rather than if you stopped walking after 1 minute is simply just NOT it all burned in your eye, yes that's true, but the VideoSense is all of the images strongly energized and each but weaker each time go search memory for a match and match the same one each time but match less each time, just like strong hobby senses income at random times like soon after wake up.

Algorithm must include weak to very strong rank neurons, weakening neurons(weakens faster the more weaker it is), deincrementing, bad rewards and label, and the stronger the neuron the faster it capacitates back to full charge and fires and incomes again.

[LOCKSUIT]

Can't we together form some processing way to see if my/your theory will work. I have it alllll, there should be a way to have it realized already than instead of me ageing every day not going to EVER get anyone to realize I have it the whooole time!

[keghn]

 Sure. You finance it and i will code/build it. Here is my synopsis of my theory. It is more detailed then yours:

https://groups.google.com/forum/#!topic/artificial-general-intelligence/UVUZ93Zep6Y


https://groups.google.com/forum/#!forum/artificial-general-intelligence

[LOCKSUIT]

@keghn

"pointers"
Pointers is sensory input searching Memory?

"There are are many types of consciousness pointers."
Before I got my theory down pat I said "there is many types of analyze senses that income/go search" In the end it was really senses labeled +/- and linked to more senses and they play out instead of magically solving things like you said.

"many pointers active like riding a skate board, and talking on the phone, and smoking a cigarette"
I have written in my theory exactly that and is as a slope like this > "/" is all of what you are doing ex. in 3 minutes PLUS multiple sensory intake like smoking AND skateboarding both pointer(s) go to look, right? !!!!!!!!!!!!!!!!!!!!!!!!!!!!!

"3D"
Thinking in 3D is clips you've seen in the past of your life. Also, crops cut out of them by them being rewards like edge/color recognition can be placed overtop another memory image that's automatically incoming for a moment ex. a nail cut-out onto a plate image. AIs in the brain build dreams.

"fly around and talk and Learn to do jobs"
A little overboard here ay hehe!

[Korrelan]

Hi BF33

Something has been bugging me about your theory, I think I have finally figured out what you are trying to describe.

Imagine a long horizontal growing memory bar/stick, divided into two rows horizontally.

First, Guess happens - actions for all the motors are generated and saved into the first box in the bottom row.

Then the actions are sent to motors.

Then senses go through rewards and are saved into the first top row box of the memory bar, above just tried actions. The senses were labelled and ranked by if they matched +/- rewards and how much they matched the reward.

The reason no one understands your theory is the way you are describing the system, your terminology is just wrong, and very, very miss leading. You also have a poor understanding of neural anatomy, programming and past/ current AI research.

Memory Front – I think your describing the state of mind in the moment or the current frame of consciousness.

Match box – This is a parallel connectionist neural net. That takes the individual sensory inputs (pixels from retina, etc) and recognises the individual pattern frames as a set of synaptic weights… in parallel. So each vision pixel, touch area value, etc is fed into the net individually but at the same time… in parallel. The best resulting pattern match is then reflected/ moved to the ‘memory front’.  The ‘strength’ of the recognised pattern is then merged with new sensory data and sent back to the neural net for re-processing.
 
+/- Rewards – This is the altering of the weights on the synapse inputs to the above ‘match box’.  You’re describing the tuning of the parallel net to recognise facets/ patterns of the incoming sensory streams.

Guess or random – Your describing the initial weights of synapse in the ‘match box’ parallel neural net.  If no exact pattern is returned then your system injects random synaptic values which generates a random action, which may or may not be recognized in the next frame.

So for simple OCR it would look like this...



If I'm right, what you are describing is a standard parallel connectionist recurrent neural model.  Many companies are already working on this type of system and have been for many years, and is in fact the basic functionality of the current ‘Deep Learning’ craze.

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

[LOCKSUIT]

BUT I have many more keys like this one:

1. Firstly +/- rewards can be man/woman fries/sour.

2. And why you laugh at NON-direct rewards like a joke or scene. The real reward has been linked to a neutral (non-rewarding sense) like shoot all planes or girl has cancer beware or laugh when gets game gift or gravity is true/false/true/false.

[LOCKSUIT]

@keghn read my reply above in case you didn't.

[keghn]

 There are allot of replies above. Just give a slimmed down repeat.

[LOCKSUIT]

"pointers"
Pointers is sensory input searching Memory?

"There are are many types of consciousness pointers."
Before I got my theory down pat I said "there is many types of analyze senses that income/go search" In the end it was really senses labeled +/- and linked to more senses and they play out instead of magically solving things like you said.

"many pointers active like riding a skate board, and talking on the phone, and smoking a cigarette"
I have written in my theory exactly that and is as a slope like this > "/" is all of what you are doing ex. in 3 minutes PLUS multiple sensory intake like smoking AND skateboarding both pointer(s) go to look, right? !!!!!!!!!!!!!!!!!!!!!!!!!!!!!

"3D"
Thinking in 3D is clips you've seen in the past of your life. Also, crops cut out of them by them being rewards like edge/color recognition can be placed overtop another memory image that's automatically incoming for a moment ex. a nail cut-out onto a plate image. AIs in the brain build dreams.

"fly around and talk and Learn to do jobs"
A little overboard here ay hehe!

[keghn]

 My AGI encodes the world into engram pattern loops.  Kinda like frames of video.
 Simple ones,  at first, and then mash them up in parallel later
on.

  I have pointers that index through the encoded engram pattern loops main memory map.

 The 3D simulator can stack many memories on top of each other. Edit in an out from past images like GIMP
image editor. At the bottom of my engram images are motor positions and audio recordings so the editor can
use them too.

 There are instruction set for cutting in and out data from the past. Croping ike you said, into a working memory area.
 The AGI randomly play with instruction set for creating existing engrams in different way and then
when it has master this it creates possible new futures later on, in the next phase.
 This is a trial and error procedure. And editing instruction set will be placed back into engram path way for re use or
re modification next time around.

 When it try out new constructions engrams, some will work and some will fail.
 The one one that fail will be its dream when it sleep. Just before they are deleted forever and a day.

[LOCKSUIT]

Ok now that that's over and done with. I leave you all to think about this below. I'll be picketing this possibly. Why? Well, look at it. This will prove I have huge stuff no-one has.
------------------------------------------------------------------------------
When you say you love steak, it is a nerve signal.

When you say you hate broccoli, it is a nerve signal.

You love one sense, and hate the other sense.

Both senses are nerve signals, and neither is actually tasty or painful. One signal is not superior than the other signal.

What happens is you have pre-installed +/- reaction-rewards from evolution, and they save generated actions when you get them. And do/stop the actions when confront again.

We have reaction-rewards to things that kept us alive - steak, and healthy women.

My point is that nothing is beautiful or tasty nor painful or disgusting, and that this is the key to knowing you are on the path to creating AI - intelligence.

Games/knowledge/movies/music too is neither unique/"hot" or disgusting. They are neutral senses not +/- rewards. They get linked to our real rewards like when you ate food and mom said "good". Then says shoot the planes "good". Music sounds match same memory sounds which are linked to culturally acquainted senses ex. power/Indian/relaxing in a jazz bar and they are linked to +rewards OR matches memory themselves and match science visions/etc. and if you love/hate such then the backpropagation to memory front just sensed will initiate & it'll be linked to ex. +rewards.

[keghn]

The AI has primitive needs. Energy, regenerating building material and the conservation of energy.
Energy is monitored at real time. like a ekg signal or stock signal. When there is a positive movement in the
signal going up in the positive direction, then this is a reward. Given by the judging part of the system. This is hard wired into the system the logic.
This hard wired reward is recorded into the engram at the time issued.

 Also, The engram path that are used are counted. When a certain path engram path has a high count a negative reward
will be issued and recorded into the engram path ways. This is to keep the system from falling into a never ending same ole
pattern loop forever.
 Consciousness pointer select path that will lead it to reward and steer away from hitting anti reward activation.

  As for human foods. The body wants food that are high in calories and protein. And foods that make you
lazy. like junk food.
 The body is very interested in gaining energy, or using as littlest energy to gain the most amount of energy.
 For foods that have vitamins, that will make you very active and cause you to burn lots of calories and are low in calories,
are vegetables and fruit. The body will only want theses when it really needs vitamins.
 
  Energy management is really big deal deal in any life form.

  The AGI will have this same problem as humans.