Time whizzes by and I, I write of glimpses I steal

Sunday, December 11, 2005

Artificial Life

A Frankenstein figure lifted me off my feet, half-a-metre from the ground and looked into my eyes savagely and said, You! Dumb Ass. Can't you even play the basic level in a car game and you call yourself intelligent. I woke up with a sweat. Wondered if I have been watching too many Hollywood movies, aliens, predators, terminators. Well! Not exactly. I had the opportunity of attending lectures by two giants in the realm of Artificial Life.

Dr.David Goldberg of University of Illinois is considered the Michael Corleone of Genetic Algorithms. (He is so eminent that I have heard of him despite not working on anything even remotely GA). Genetic algorithms are a particular class of evolutionary algorithms that use techniques inspired by evolutionary biology such as inheritance, mutation, natural selection, and crossover.

David and his team have come up with the next generation of Genetic algorithms (basically because he has been able to increase the speed of computation about 173 times). Though I must admit that some of his lecture went right over my head, it wasn't all too bad. I especially liked his use of 'supermultiplicity'. Well! Now that is SUPERINTELLIGENT.

The second speaker, Dr. Mark Bedau of ProtoLife was no less brilliant. His area of research is the creating of Artificial cells in a lab. Wet life, he said, as against soft life which is where GA's come into play and hard life where Robots come into the picture. It goes beyond mere cloning or gene-manipulation. The whole cell is designed and manufactured in a lab with available chemicals and possibly with some MEMS (which is Micro Electro-Mechanical Systems). Science Fiction. Nope! Research of the future.

On the whole, an eye-opener. To the infinte possibilities of life. Natural or artificial.

3 comments:

romantic by twilight said...

interesting...

Nosophorus said...

Hallo, Karthik!

"I had the opportunity of attending lectures by two giants in the realm of Artificial Life."

Well, if you had said names such as Chris Langton or Karl Sims, I would agree with you. Genetic Algorithms and Artificial Life are different research areas.

"Dr.David Goldberg of University of Illinois is considered the Michael Corleone of Genetic Algorithms."

I completely agree with the comparison! Anyone who disagrees with Dr Goldberg Corleone is "killed" by his prima donna sensibility which does not accept any type of critiques toward his opinions/research/work.

I know what I am saying, since I am forbidden to post in his blog. It's just a shadow of Dr Goldberg Corleone's prima donna sensibility.

You should asking yourself: "Why that guy is saying all those words?". Well, here you are some links (read them in order, please):

1 - http://illigal.blogspot.com/2006/02/selling.html

2 - http://illigal.blogspot.com/2006/02/change-world.html

3 - http://illigal.blogspot.com/2006/05/hboa-patent-issues.html

4 - http://illigal.blogspot.com/2006/05/in-defense-of-patents.html

"David and his team have come up with the next generation of Genetic algorithms (basically because he has been able to increase the speed of computation about 173 times)."

But I consider "modern" "Genetic" Algorithms lost their biologic inspiration, see hBOA, where "traditional" evolutionary operators are changed by probabilistic sampling. We had (and have) so many problems related to "old school" GAs, such as the rotation of the coordinate system that results in a drastic performance loss of GAs. Theoretical considerations seem to indicate that a full rotation, which makes each parameter dependent on each other, results in a convergence rate that is slower than the convergence rate of random search. These (negative) indications, however, are quite a bit unpopular (see here), mainly when Corleone family members (GAs heavy enthusiasts, such as Dr Goldberg Corleone) only say what they want to say about that subject and neglect (malevolently?) other kinds of Evolutionary Computation, for instance Evolution Strategies and Evolutionary Programming. I consider that all those (negative) indications led Corleone family members (GAs enthusiasts, such as Dr Goldberg Corleone) to change the approach ("traditional" evolutionary operators) they were using and Corleone family members will not hesitate to change their approach again if necessary, principally if results got by researchers (and even by themselves) do not match with Corleone family members' GA theory (see, for example, the Buiding Block Hypothesis and the results got by Evolutionary Computation researchers around the world). Nowaday there are new and weird genetic-like methods.

Now, imagine how many new problems we should have with those newbie GAs! And they are just beginning. We must be reasoned persons and do not take into account our personal opinions and desires as the real truth. :)

Até Mais!

Marcelo

Ayisha said...

nice site for more books I have some more gifts..



Reference books

Books

Kitaben

books

Reference Books

tutorials

rapidshare tutorials

rapidshare books

MZWorld

Upload Books

MZWorld Library

Books Forum