Society of Robots - Robot Forum
Software => Software => Topic started by: lemontree on July 18, 2008, 03:07:02 AM
-
I recently found what I consider to be one of the most useful AI papers I have ever read:
http://arxiv.org/pdf/0705.1460.pdf (http://arxiv.org/pdf/0705.1460.pdf)
There is some really great information about how to use genetic algorithms to design self-assembling logic gates that can do useful things like act as a general purpose ALU or Multiplier. The possibilities are endless. The self-assembling aspect seems to be the key to evolving worthwhile systems.
Looking at it again I guess it is nearly unreadable. 1/10 for style, 10/10 for content.
If you present things badly you get no respect and are ignored, no matter how important your discovery. I do know that from experience. To save the day I have done a fact sheet for the algorithm:
http://www.freewebs.com/ssevector/assembly.htm (http://www.freewebs.com/ssevector/assembly.htm)
-
There was a link here but then I did take it away.
An Arithmetic Logic Unit (ALU) is one of the main subcomponents of a microprocessor. If you can evolve that using a genetic algorithm, then evolving the other subcomponents should also be possible. Eh, microprocessors designing microprocessors with minimal human intervention. Actually humans must design microprocessors using subcomponents in order to understand and organize what is going on.
An evolutionary system can take a more integrated approach where everything is fitted together without the conceptual need for factoring the problem into subcomponents. Undoubtedly this would yield a much more efficient design.
-
self programmable FPGAs. geesh. isn't that how would an EE abstractise a human brain? :P