Also I'm running the Face2face competition.
Well... I had some spare time lately, and also my *fast* computer crashed and it will take me some time to get a new one, so in the mean time I navigate the forums. Also, I think the Wiki is the greatest Robocode resource out there, because it is the first centralized/permanent/easily maintenable Robocode knowledge site. So I'm happy to contribute. -- Albert
True. It is a pure statistical approach (you segment the data, build the probability functions) and fire to the most probable bearing. The only difference between BPS and classical statistical approach is that BPS saves many families of probability functions (pe. by distance and heading, by distance and accumulated velocity, etc) and always uses the best for a given moment. But it nevew worked as expected... because the ammout of data you need to make all this information is so high that learns very slowly. -- Albert
About this probability functions. I'm trying to include some bayes nets in robocode. The problem is that i'm new in this and for now everithing is floating in a dense fog for me. You sed you don't want to share the source but you are willing to give some answers. Can you help me on building this probabilities? Can i contact you by email or how? --H7
Albert, is there any way that I can get into contact with you via email? I am seriously considering updating RoboRumble@Home so it will be able to run with current Robocode versions. In fact, I already did an alpha version that runs with Robocode 1.2.5A, but has some issues. Is it okay with you that I update RoboRumble@Home? Should we create a project page on SourceForge? like it's done with Robocode? Do you have any objections, restrictions, ideas or similar? :-) --Fnl