GP-Lab
The Genetic Programming Laboratory
A Windows-32 Evolutionary Programming Algorithm Generator
Current Version: 0.9 (BETA)
News
Screenshots
Download
History
To Do List
E-mail me!
|
The Genetic Programming
Laboratory
Introduction
Welcome to the web site of GP-Lab.
The Genetic Programming Laboratory is a user-centric
toolkit that is an all-inclusive system designed
with the average programmer in mind. GP-Lab's
development is directed toward a software developer
who knows little about Genetic Programming, has a
'black-box' solution in mind and the tools to solve
it, and needs to 'reverse-engineer' a program.
GP-Lab allows the user to enter all functions,
tools, and general parameters into the system
through a user-friendly series of data entry panels.
GP-Lab started in November of 2003 as an extension
of Mike Kramer's 1996 work, GAPS:
Genetic Algorithm Programming
System. GP-Lab is a second incarnation of that
project, incorporating a full procedural programming
language in the style of C or Pascal, known as "A
Genetic Control Teletype", or AGCT for short.
How It
Differs from Other GP Tools
GP-Lab has several features not available in all
GP toolkits:
- The use of a pervasive procedural
language defining the base program/ADF,
user-defined fitness calculations,
and user-defined
functions and procedures
- Automatic code generation into C++ of
'honor roll' individuals
- The ability to view individuals in in
either functional ("LISP-like") code, or in
tree format.
News
- 02Nov2004 Completed Download section,
added e-mail address (sorry for not adding that earlier!)
Beginning the Power Point Presentation for the IEEE Conference.
Should have that done by the end of this week.
- 03Sep2004 GP-Lab Website
initiated and published.
- 17Aug2004 GP-Lab conference
paper is accepted at the IEEE Artificial
Tools. I will be presenting the paper
personally at the conference in November in
Boca Rotan, Florida! I'm pretty excited
about this turn of events. One of the
paper's readers would like to use GP-Lab as
a teaching tool in his graduate AI course.
- 28Jun2004 GP-Lab conference
paper edited by Dr. Du
Zhang of the Engineering and Computer
Science department at the California State
University at Sacramento (CSUS) and
submitted to the sponsors of the 16th
IEEE International Conference on Tools with
Artificial Intelligence.
- 15May2004 GP-Lab Master's
Project submitted for graduation in
Software Engineering from CSUS's Engineering
and Computer Science department
- 05May2004 GP-Lab Version
0.9/Beta completed for submission for
graduation of the Master's Project.
- 17Oct2003 Beginning work of
GP-Lab initiated
History
Future Features (A.K.A.
'To-Do')
The following is a list of major items planned for
GP-Lab's future:
- Implementation of 'Debug' Mode -
One of the main things missing in GP-Lab is
a debug mode in which the user may set up
'watches' to view attribute tables and
function return values during and
post-execution
- Implementation of importing
Lil-GP environment files - The feature
will allow a user to implement environment
configuration files developed in 'lil-gp,'
the common, popular unix-based GP development
kernel.
- User-defined random seeds -
Currently, GP-Lab is designed to use a
Merseinne-Twister random number seeded
purely by date-time functions, which
ensures that each execution is completely
random and unlike any previous execution.
The implementation of a user-defined
random-number seed will ensure that every
execution will be precisely alike each
time.
- User-defined Tournament Modes -
(Credit to Dr. Scott
Gordon of CSUS) When implemented, this
feature will allow a user to select a number of
high-fitness individuals, rather than only the
absolute best one of a randomly selected
group.
- Reading data set files - This
feature will allow a user to use a file of
data for target criteria, rather than
entering it manually.
- Implementation of 'Training
Sets' - (must be implemented after the
above) This feature, when implemented,
will allow a user to use a selected
percentage of a data file as a 'training'
data set, then to use the remaining as a
'success' target set.
Contact
I can be e-mailed at
bglaholt@theglaholts.net (remove the phrase 'SPAMEX' from the address).
Alternatively, I am also available on occasion in ICQ (ICQ# is 12223860,
under the name of "Toothy/YD/Shniz")
|