The cheat part is easy to eliminate: if EI is used, then there is no program
and no cheating. Leave the stereoscopic vision out. I can watch a black and
white movie with one eye and still recognize Elyzabeth Taylor. I can even
watch a black and white picture of her, no grays, with one eye and recognize
her. So I propose a minimalist model that abstracts only the causal relation
light=>signal and the adjacency relation between pixels, nothing else.  

I am not saying this is it. I am saying starting with this, the same I
already did, just larger scale and better automation. Then grow. It is a
necessary first step anyway. It would help to debug the program that does
the IO and minimizes the functional, if anything else. 

Sergio


-----Original Message-----
From: Alan Grimes [mailto:[email protected]] 
Sent: Monday, July 16, 2012 1:09 PM
To: AGI
Subject: Re: [agi] Analog Computation

Sergio Pissanetzky wrote:
> Alan,
> That's nonsense. A camera, or a person, can look at a black and white 
> scene, light-on or light-off. A person will still recognize the image.
How?

Sorry, the simplest vision I'm willing to consider is 8-bit gray-scale.
The 1-bit monochrome that you describe is too simple and makes it possible
to Cheat. Furthermore, it doesn't really help because line art is actually
more difficult to draw and interpret because the 3D information needs to be
inferred from experience.



--
E T F
N H E
D E D

Powers are not rights.





-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/18883996-f0d58d57
Modify Your Subscription:
https://www.listbox.com/member/?&;
d2
Powered by Listbox: http://www.listbox.com





-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968
Powered by Listbox: http://www.listbox.com

Reply via email to