RE: genetic programming
Genetic algorithms are interesting and I just finished
going over "Genetic Algorithms + Data Structures = Evolution Programs"
Third edition by Michalewicz published by Springer-Verlag.
Although not from MIT my PhD is from IIT in Computer Science and
a powerful PC can do a lot with Genetic Algorithms. However, I
don't see much application to TC development. As I previously
mentioned, our empirical data is rapidly increasing and when
there is enough, a hybrid model combining fuzzy logic and existing
circuit theory seems to be a reasonable solution. These solutions
are not perfect but work in acceptable boundaries. If you want to
discuss this issue further, contact me individually. This discussion is
complex and would generate many off topic, close topic, and
nearly close topic issues. (See how nice linguistic hedges work!).
John W. Gudenas, Ph.D.
Department Chair of Computer Science and Mathematics
Aurora University, Aurora IL 60506
Office: 630-844-5539 Fax: 630-844-7830
From: Tesla List [SMTP:tesla-at-pupman-dot-com]
Sent: Saturday, July 10, 1999 11:05 AM
Subject: genetic programming
Original Poster: BAREN01-at-aol-dot-com
I was watching a video that MIT sent me (to advertise thier new book
"genetic programming III") where they devised a way to use genetic programmig
to design circuitry. The impressive thing was that they had used this
program to design circuitry that infringed on, and in some cases improved
earlier patents on circuitry, such as high pass filters and low pass filters.
It was all very intersting, and it occured that this could be very
usefull in designing primary circuitry, or power supply. I was wondering if
any one had ever tried this or concidered this. I doubt that any one has,
unless they hapen to go to MIT or U of I. Particularly because most of the
problems were run of a "beawolf cluster computer" that operates at 5 gigahz,
which I doubt any one on the list has (price tag $15,000). But hey, it's
worth checking, hey, maybe someone's into programming.