Ai Forums Home Welcome Guest    Friday, November 24, 2017
Ai Site > Ai Forums > The Artificial Intelligence Forum > Prof. Yiannis Aloimonos | A minimalist ... Last PostsLoginRegisterWhy Register
Topic: Prof. Yiannis Aloimonos | A minimalist grammar of human action

keghn
posted 8/4/2016  05:47Send e-mail to userReply with quote
Prof. Yiannis Aloimonos | A minimalist grammar of human action:

https://www.youtube.com/watch?v=vKEhcxGUp18

 Prof. Yiannis Aloimonos | A minimalist grammar of human action

useruser
posted 8/5/2016  00:12Reply with quote
Makes you realize that a working A.I. robot now is almost like a bunch of people in the 18th century discussing what would be required for a trip to the moon to become possible in the next 50 years. I think we will get there but we are like these 18th century dreamers-it will happen but not in our lifetime.


rouncer
posted 8/7/2016  11:57Send e-mail to userReply with quote
[EDIT] I actually watched the video, it looks pretty good!!! WHAT ARE U WORRIED ABOUT?!!?? of course itll happen. [/EDIT]

yes, there is a lot of dreamers. but there is guys that actually know how to do it now.

heres a VERY important thing ill tell you now, but its not complicated.

If u need a robot straight away, without too much difficulty there is a kind of neural network which you can write in about 5 or so minutes, start to finish, and its actually the quickest, but it doesnt handle near neighbour (which is for classing - GROUPING PATTERNS). but it can be advanced to have it, if you need it.


What I would like you to do, is go allocate the largest amount of ram in one array you can.

The array is INPUT PATTERN SIZE * OUTPUT PATTERN SIZE.

u then directly access the output from the input, and now you have a content addressable memory for up to about 28 bit key length, in about a gigabyte, depending on how many output bits you needed.

CONVERTING I TO O IS VERY IMPORTANT!

"I see this, so I do this."

"This causes this reaction".

INPUT -> OUTPUT

to add near neighbour in such an easy fashion, to get it to access as quick as an exact match, it isnt what you would call resourcefully modest. If you repeat that array per bit of the input key, u can slip off the first array onto the second array, and then so on and so on, and get the nearest distance. but u must store in every single array to get a correct assumption.

And u can also write it for approx distance.

There you go, take it or leave it. - but what if you didnt understand? Thats why us guys that actually know what we are doing are hidden in a sea of bullshit artists. :)
Trust or dont. But there is no ego or rival making in this, that matters a damn.

Get the job done, is the important thing.

Its not far from here, about 4 or so of these hooked up with a few ordinary filters is the whole thing.
And it does two things, make a copy of its environment, to do an isotropic search.
Then go take it out to a farm and herd sheep with it! hehehehhee

Last edited by rouncer @ 8/7/2016 12:12:00 PM

AiHasBeenSolved
posted 8/7/2016  16:46Send e-mail to userReply with quote
Anything to advance the State of the Art in Artificial General Intelligence.

 State of the Art in Artificial General Intelligence

rouncer
posted 8/7/2016  17:28Send e-mail to userReply with quote
Yes, indeed AiHasBeenSolved. (cool name.)

heres a bit of programmin I did today on my topic->
https://www.youtube.com/watch?v=DqxN91EKY4Y&feature=youtu.be

just about ready for testing - sans the actual motor gen. (but its not that hard. but i havent actually ironed it out in theory exactly how im going to go about it but its as good as done dudes.)
https://youtu.be/ayphdOTFJts

Last edited by rouncer @ 8/7/2016 8:58:00 PM
  1  
'Send Send email to user    Reply with quote Reply with quote    Edit message Edit message

Forums Home    Hal and other child machines    Alan and other chatbots    Language Mind and Consciousness  
Contact Us Terms of Use