Re: Open problems in GA/ES/EP/GP

Quaternion wrote:
> 
> Hello,
> I am a student who would like to write a thesis on a subject in the field of
> Evolutionary Programming, in the coming two years. However I am unsure what
> problems or studies are still somewhat 'open'. I would have liked for
> instance to research the image compression technique using fractals,
> implemented with GA's as optimization modifications; however I 'fear' image
> compression has already been thoroughly investigated. I would also be
> interested in the application of GP for encryption - and decryption
> possibilities, or more practically oriented, the designing of hardware
> specifically designed for creating programs derived from GP and sets of
> parameters. Especially the latter seems very interesting for me since it is
> not entirely my field, yet again I think I am too late to be able to say
> anything new about it.
> Does anybody have any idea where I could go to to come up with an insured
> *original* subject to research?

I have a suggestion, which is: Neuroevolution applied toward problems
where active use of memory is required after the training period is
over.

Although artificial neural networks are said to have memory, this memory
is almost always anagous to ROM.  The information is stored in the
weights during a training period.  Afterwards the network does not store
new data during use. You could say the network has no RAM.  Or you could
say it only has long term memory, no short term memory.  Nevertheless,
it is possible to for an artificial neural network to have and use short
term memory.  This is an area where more research is desirable.

Mitchell Timin

-- 
"Many are stubborn in pursuit of the path they have chosen, few in
pursuit of the goal." - Friedrich Nietzsche

http://annevolve.sourceforge.net is what I'm into nowadays.
Humans may write to me at this address: zenguy at telus dot net