The last invention.

  • 388 Replies
  • 51332 Views
*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *************
  • Transformer
  • *
  • 2116
  • First it wiggles, then it is rewarded.
    • Enter Lair
Re: The last invention.
« Reply #345 on: July 12, 2018, 03:41:12 pm »
not a big difference though when we link the 2 together though...

that's the point / question.....
« Last Edit: July 12, 2018, 04:10:55 pm by LOCKSUIT »
Emergent

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *************
  • Transformer
  • *
  • 2116
  • First it wiggles, then it is rewarded.
    • Enter Lair
Re: The last invention.
« Reply #346 on: July 12, 2018, 04:17:14 pm »
see pic.....how do we do this?

I want to link 2 features together to create new, re-useable feature made of 2 small ones.

Our brains must already have the connections from birth, simply it takes the 2 activated and strengthens 2 connections to a the same node in a deeper layer.
Emergent

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *************
  • Transformer
  • *
  • 2116
  • First it wiggles, then it is rewarded.
    • Enter Lair
Re: The last invention.
« Reply #347 on: July 12, 2018, 04:35:40 pm »
well more like this strictly speaking
Emergent

*

ranch vermin

  • Not much time left.
  • Terminator
  • *********
  • 958
  • Its nearly time!
Re: The last invention.
« Reply #348 on: July 12, 2018, 05:28:28 pm »
What your suggesting (which i thought of AGES ago),           "reusability inside a neuron network"

If you want to trust me, (anyone who reads this.) I think it comes down to logical simplification,  yes you can simplify logic gates, but you only get a realistic amount of compression out of them,  so its 30% the size tops.   other than that, you have to store everything individually, no bits for free.   And I think thats only if the patterns belong in an OR/ in the same class / arise at the same cell.  if they are separate classes then you dont even get to do it.

I think the solution to this super robot is possibly simple,  but ur looking in the wrong place.   U NEED TO IMPLEMENT locksuit!,  OR U WONT GET ANYWHERE!!!

Stop Hijacking Korrellans thread,  he doesnt need your work here, its about HIS.

*

korrelan

  • Trusty Member
  • **********
  • Millennium Man
  • *
  • 1092
  • Look into my eyes! WOAH!
    • YouTube
Re: The last invention.
« Reply #349 on: July 13, 2018, 02:40:30 pm »
Quote
see pic.....how do we do this?

I want to link 2 features together to create new, re-useable feature made of 2 small ones.

The simple answer is… we don’t.

Your brain is not using single words/ nodes and links, you are thinking about the problem space at too high an abstraction level.

It’s like the lines you draw and words/ letters you type don’t exist inside your computer; they exist purely as interpreted binary code… It’s similar with your brain… There are no simple nodes and links in the way you perceive them…

You have 80 billion ish neurons and each neuron can have thousands of synapse, that’s approximately a quadrillion synapse… that’s a lot of synapse.  Every thought ‘frame’ you have heightens neural activity in approximately 20% of your brain at any one time, though most of the brain is always active.  Different thoughts will use different areas, so even working at the conservative estimate of 20% that’s about 4 billion neurons and 4 trillion synapses just to think about a cup, or look out the window, or scratch your nose… not a single connection… but trillions.

You exist in your own personal simulation of reality and it’s different to everyone else’s, everything you experience comes from your senses, even colour doesn’t exist in actual reality… it’s just a construct of your brain like everything else.

When you think about a cup you are mentally modelling every aspect of the cup, what it looks like, what its made of, what it’s uses are, etc.  You are mentally modelling a concept with properties and sub properties, not a single simple node/ word.

You asked how your brain links two memories/ concepts; you have to consider the complexity and depth of the schema that is doing the ‘linking’, it’s not as simple as drawing a line between two nodes/ words,.. I wish it was.

 :)
It thunk... therefore it is!

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *************
  • Transformer
  • *
  • 2116
  • First it wiggles, then it is rewarded.
    • Enter Lair
Re: The last invention.
« Reply #350 on: July 13, 2018, 03:41:39 pm »
Is it true that a hidden node in the middle of a CNN or GAN can reverse down to the leaves and reconstruct a eyeball or face or nose? What about those patches on black on this page at the very top (put your mouse on one of them)? https://distill.pub/2018/building-blocks/

You mean my brain is using weights/representations right? Not any sort of actual word/object...1 word is really a fuzzy cluster of dots like a dandelion. And it's lighting up all the connected features that link to cup like container, item, cheap, round, hard, coffee, drink. Maybe hard cheap drinkable etc link together to a representative node meaning cup. Wouldn't it be great though if it had a title "cup" on the node?
I think so. They don't make up the word cup, (or wait maybe they do, while all these minus hard make up hard, each making up each other (needing a few more word properties for others since not all words have same properties)), rather each is a separate word with properties linked and each can be used to create bigger features.

Well if 1 neuron/column did hold 1 word or audio snippet, it would work like a word just what is stored is like binary on a disc it is weird 'being' saved.

Ok so you're saying I'm not actually linking cup and socks but instead the concepts ex. I'm linking the cluster "drink, coffee, container, hard" to "soft, feet, cuddly, warm, item"? Isn't that a explosive controduction of the problem in the first place!? Really we just link the 2 dandelions together by linking cup to socks...

Is it true that the brain is making representation nodes that stand for both cups/bottles/etc, not a 'word' like container but it could be thought of being like a word, like "container", which'd stands for cup/bottle/bucket/etc that hold liquid/etc.

Ok great a representation is made but wouldn't it be awesome for it to have its proper name, container ! ?
Emergent

*

korrelan

  • Trusty Member
  • **********
  • Millennium Man
  • *
  • 1092
  • Look into my eyes! WOAH!
    • YouTube
Re: The last invention.
« Reply #351 on: July 13, 2018, 11:15:06 pm »
Quote
Ok great a representation is made but wouldn't it be awesome for it to have its proper name, container ! ?

You asked me my opinion/ theory on how the human brain stores and links memories.

That doesn’t mean this is the only method or even required, there may be many ways to achieve an AGI.

 :)

It thunk... therefore it is!

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *************
  • Transformer
  • *
  • 2116
  • First it wiggles, then it is rewarded.
    • Enter Lair
Re: The last invention.
« Reply #352 on: July 24, 2018, 01:02:35 pm »
Is the frontal cortex the facts describing facts area here we're talking about? I see now!
Emergent

*

korrelan

  • Trusty Member
  • **********
  • Millennium Man
  • *
  • 1092
  • Look into my eyes! WOAH!
    • YouTube
Re: The last invention.
« Reply #353 on: July 24, 2018, 06:50:23 pm »
Quote
Is the frontal cortex the facts describing facts area here we're talking about?

Yeah… sort of…the prefrontal cortex is the largest swathe of cerebral cortex that has no sensory input/ outputs… so its functionality is purely derived from the rest of the cortex.

It’s the largest bit dedicated to learning what the rest of the cortex is doing, though there are many smaller areas evenly distributed across the cortex that do a similar jobs relevant to their adjacent cortex areas.

 :)
It thunk... therefore it is!

*

korrelan

  • Trusty Member
  • **********
  • Millennium Man
  • *
  • 1092
  • Look into my eyes! WOAH!
    • YouTube
Re: The last invention.
« Reply #354 on: August 15, 2018, 09:51:26 pm »
You have probably heard about the persistence of vision but what about the persistence of motion? 

During my AGI research into the persistence of vision I realised that my system is also exhibiting a kind of persistence of motion.  I needed some way to test if humans actually do experience this phenomenon.

I eventually came up with an experiment that anyone can try.

Simply stare at the centre of this video for a 10 seconds (starting from 1:03 in video), and then look away at a picture on the wall, or your keyboard… any fixed stationary object.

WARNING… DO NOT DO THIS… If you suffer from epilepsy or any other sensory/ frequency triggered metal problems.



Did you see the persistence of motion?

 :)
It thunk... therefore it is!

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *************
  • Transformer
  • *
  • 2116
  • First it wiggles, then it is rewarded.
    • Enter Lair
Re: The last invention.
« Reply #355 on: August 15, 2018, 10:17:43 pm »
All I see after looking away at my wall is a triangle at the right side. No worky...

Nonetheless I don't think it amounts to anything important in the AGI...explain if it does...
Emergent

*

ivan.moony

  • Trusty Member
  • **********
  • Millennium Man
  • *
  • 1109
    • Some of my projects
Re: The last invention.
« Reply #356 on: August 15, 2018, 10:50:06 pm »
It works, static room wall seems like going away...

I noticed this kind of effect when looking to the back window while driving in a moving bus, and when the bus stops, the effect starts. I also saw experiments on Facebook or similar pages that were created specifically to induce such effects. Some pretty mean things are possible, like looking your room swirls in and out in multiple centers of rotation after watching specific 5 minutes video. The effect lasted for 30 seconds, but it was very cool.
Dream big. The bigger the dream is, the more beautiful place the world becomes.

*

korrelan

  • Trusty Member
  • **********
  • Millennium Man
  • *
  • 1092
  • Look into my eyes! WOAH!
    • YouTube
Re: The last invention.
« Reply #357 on: August 15, 2018, 11:06:53 pm »
Ah I see they call it the motion after-effect, not the persistence of motion… I’d never heard of this before… sweet.

 :)

@Lock

It shows that the visual mechanism sensing motion has a kind of inertia, and motion is not sensed/ calculated on a frame to frame basis.  So the visual cortex neurons representing sensed motion at specific regions in the visual field are firing for longer durations dependent on the length of stimulation… probably not relevant to your AGI.

 :)
It thunk... therefore it is!

*

korrelan

  • Trusty Member
  • **********
  • Millennium Man
  • *
  • 1092
  • Look into my eyes! WOAH!
    • YouTube
Re: The last invention.
« Reply #358 on: September 25, 2018, 12:20:16 pm »
I was prompted by the MIT article posted a few days ago about object occlusion to dig out an old video from when I was initially designing my AGI schema.

http://aidreams.co.uk/forum/index.php?topic=13408.msg54867;topicseen#new

Software is running on a single core 2.4Ghz PC instantly recognising objects with various rotational/ scale invariance’s and occlusions from a single quick viewing/ mapping.  The system only has to view the object once and is then able to recognise any object, at any size or angle, even if partially occluded.

Like the rest of my project this technique is based on the human visual system.



 :)
« Last Edit: September 25, 2018, 12:46:46 pm by korrelan »
It thunk... therefore it is!

*

ranch vermin

  • Not much time left.
  • Terminator
  • *********
  • 958
  • Its nearly time!
Re: The last invention.
« Reply #359 on: September 25, 2018, 03:25:25 pm »
Thats cool.

Worked just how you said.

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

 


redoing my site
by ruebot (General Project Discussion)
Today at 02:25:34 am
How about VR, AR, MR? What?
by Art (Virtual Reality)
Today at 01:36:46 am
ChatbotML on Twitch
by 8pla.net (Home Made Robots)
October 21, 2018, 07:28:02 pm
Hello there !
by Freddy (New Users Please Post Here)
October 21, 2018, 03:55:35 pm
Cats are evolving...
by DemonRaven (General Chat)
October 21, 2018, 08:27:03 am
With great power, comes great laziness.
by DemonRaven (General AI Discussion)
October 21, 2018, 07:59:01 am
ETHICS
by DemonRaven (General AI Discussion)
October 21, 2018, 07:47:43 am
A most unique coin
by korrelan (General Chat)
October 20, 2018, 06:12:06 pm

Users Online

96 Guests, 0 Users

Most Online Today: 117. Most Online Ever: 208 (August 27, 2008, 09:36:30 am)

Articles