JGAP 3.4 released (Java Genetic Algorithms Package)


News: JGAP 3.4 released (Java Genetic Algorithms Package)

  1. JGAP is a sophisticated Genetic Algorithms and Genetic Programming package written in Java. JGAP version 3.4 is an enhancement release, driving Genetic Programming capabilities to a new level. Among the extensions are: * A new popular example: Mona Lisa Painting Problem, available for GA as well as for GP * Enhancement of evolution cycle of Genetic Programming * Javadoc-enhancements and new JUnit tests * Some bug fixes This release can be downloaded here: http://sourceforge.net/project/showfiles.php?group_id=11618&package_id=48940 For more information visit the JGAP homepage at http://jgap.sf.net Klaus Meffert for the JGAP team
  2. One word: AWESOME! Now, can i use this to help me out with the totally FUCKED UP captcha system on theserverside? I have to reload 3 times in avg to get a picture i can decode. Very annoying.
  3. I am interested in JGAP, but I find that release notices seem more appropriate coming from somewhere like freshmeat.net. Can I recommend that you only announce major versions (in the past, you've even posted a release candidate here,) and that you provide some rationale for considering JGAP enterprise software. Perhaps you can include some examples in the software that are more enterprise-oriented than painting the Mona Lisa. You could also have links from your sites to enterprise-oriented applications of GP. I have written two GA applications based on JGAP. Each time, I felt constrained by the JGAP API and rewrote the application using a different framework (Evolvica and Watchmaker.) The downside of Evolvica is that the project is defunct.
  4. In the internation timebabling competition http://www.cs.qub.ac.uk/itc2007/ tabu search algorithms seem to beat genetic algorithms - for those planning problems at least. Has anyone tried a JGap implementation of those specs? Drools Solver - which I created (yes, this is a shameless plug) - uses a tabu search / simulated annealing algorithm on top of the drools rule engine for score calculation. I suspect genetic algorithms too, could delegate the score calculation to a rule engine too. This makes it a lot easier to add score constraints for a small scalable cost. You can download Drools Solver and it's examples here: http://jboss.org/drools/downloads.html The Drools Solver manual is here: http://users.telenet.be/geoffrey/tmp/solver/manual/html_single/