Overview
Examples
Screenshots
Comparisons
Applications
Download
Documentation
Tutorials
Bazaar
Status & Roadmap
FAQ
Authors & License
Forums
Funding Ultimate++
Search on this site
Search in forums












SourceForge.net Logo
Home » Community » Coffee corner » My new OSS project...
My new OSS project... [message #24172] Thu, 31 December 2009 13:30 Go to next message
mirek is currently offline  mirek
Messages: 13975
Registered: November 2005
Ultimate Member
I have registered a new project on GoogleCode:

http://code.google.com/p/genai/

It is supposed to be the platform for experimenting with "evolving" or "selfemerging" AI. Most likely, it will lead nowhere, but I am quite a believer in possibility of Strong AI in near future (20-30 years) and I would like to try some things myself...

Of course, the code is for U++ Smile

There is so far only one interesting package, "Gen3", and in fact, it is more funny than anything else.

It tries to "evolve" algorithm for white dots that are moving and eating green dots... (And to make it more fun, sometimes some white dots convert to red dot predators, eating white dots )

(Note: the basic result/failure of experiment is that dots do not seem to evolve reactions on sensory input. Instead, they do evolve very complex movement patterns that seem to replace the need for sensory inputs. Well, uhm, I suppose that is how evolution is supposed to work, after all)

In future, I am planning on "Pac-Man" experiment - trying to find some ways to implement AI capable to evolve engine capable learning to play various variants of the game. We will see...

Mirek

[Updated on: Thu, 31 December 2009 13:32]

Report message to a moderator

Re: My new OSS project... [message #24177 is a reply to message #24172] Thu, 31 December 2009 18:19 Go to previous messageGo to next message
andrei_natanael is currently offline  andrei_natanael
Messages: 262
Registered: January 2009
Experienced Member
Hi Mirek,
I've gave a try to genetic algorithms but didn't get it to the end(for a project at university), I hope now i will have the time and patient to watch this project evolving Razz learn something from it and why not writing code if necessary.
Re: My new OSS project... [message #24181 is a reply to message #24177] Fri, 01 January 2010 12:00 Go to previous messageGo to next message
koldo is currently offline  koldo
Messages: 3355
Registered: August 2008
Senior Veteran
Hello Mirek

Curious and interesting effort. I did not know you where interested in AI.

Best regards
Koldo


Best regards
IƱaki
Re: My new OSS project... [message #24183 is a reply to message #24181] Fri, 01 January 2010 12:36 Go to previous messageGo to next message
mirek is currently offline  mirek
Messages: 13975
Registered: November 2005
Ultimate Member
koldo wrote on Fri, 01 January 2010 06:00

Hello Mirek

Curious and interesting effort. I did not know you where interested in AI.



Is there any computer geek who is not? Smile

Re: My new OSS project... [message #24192 is a reply to message #24172] Mon, 04 January 2010 09:11 Go to previous messageGo to next message
mr_ped is currently offline  mr_ped
Messages: 825
Registered: November 2005
Location: Czech Republic - Praha
Experienced Contributor
"... but I am quite a believer in possibility of Strong AI in near future (20-30 years) and I would like to try some things myself..."

If possible at all, then IMO yes, it may happen.
Although I'm afraid the 20-30 years is a mix between speed of HW (exponential) and SW (limited by smart humans adding small pieces every year) evolution, and the strong AI itself will be more like pure SW.
But this thing may be tricky to predict, once somebody will jump on the right track, the next development may be blazingly fast, like 10 years from that point and the AI will be very strong. And there may be as well as 100 years of dead ends before that point.

Although I'm not very current with recent AI development, so I may be well off the real status.
Re: My new OSS project... [message #24193 is a reply to message #24192] Mon, 04 January 2010 09:32 Go to previous messageGo to next message
mirek is currently offline  mirek
Messages: 13975
Registered: November 2005
Ultimate Member
mr_ped wrote on Mon, 04 January 2010 03:11

"... but I am quite a believer in possibility of Strong AI in near future (20-30 years) and I would like to try some things myself..."

If possible at all, then IMO yes, it may happen.
Although I'm afraid the 20-30 years is a mix between speed of HW (exponential) and SW (limited by smart humans adding small pieces every year) evolution, and the strong AI itself will be more like pure SW.



Well, what I would like to try is to use sort of genetic algorithm to develop "seed" AI SW. That might shortcurcuit those 100 years of dead ends...
Re: My new OSS project... [message #24194 is a reply to message #24193] Mon, 04 January 2010 11:00 Go to previous messageGo to next message
mrjt is currently offline  mrjt
Messages: 705
Registered: March 2007
Location: London
Contributor
Very interesting. I guess most programmers have wondered about the possibility of this at some point, I know I have.

The main problem that occurred to me was creating a complex enough environment to make sensory input useful to the algorithm, which is the problem I think you've encountered. Additionally the AI may need a matching complexity in it's interactions (or influence) with it's environment.

After all, in the natural world you have to get all the way up to very complex multi-celled organisms before you even get light sensitive skin patches.
Re: My new OSS project... [message #24198 is a reply to message #24194] Mon, 04 January 2010 12:42 Go to previous message
mirek is currently offline  mirek
Messages: 13975
Registered: November 2005
Ultimate Member
mrjt wrote on Mon, 04 January 2010 05:00

Very interesting. I guess most programmers have wondered about the possibility of this at some point, I know I have.

The main problem that occurred to me was creating a complex enough environment to make sensory input useful to the algorithm, which is the problem I think you've encountered. Additionally the AI may need a matching complexity in it's interactions (or influence) with it's environment.



Yes. Exactly.

Quote:


After all, in the natural world you have to get all the way up to very complex multi-celled organisms before you even get light sensitive skin patches.



Well, I would say that with HW power big enough, you should in theory be able to reconstruct the whole evolution.

The really tricky part is how to cheat and what to exclude Smile
Previous Topic: Something relevant...
Next Topic: Newbie Corner
Goto Forum:
  


Current Time: Thu Mar 28 22:48:14 CET 2024

Total time taken to generate the page: 0.01810 seconds