JavaFX ScriptBots
125 views
Skip to first unread message
Skid Runner
May 4, 2015, 4:28:23 PM5/4/15
to scrip...@googlegroups.com
I have forked scriptbots and plan on making a JavaFX version while I agree with using c++ for speed and efficiency I am a Java developer so I would like to see the performance difference between using OpenGL and C++ and JavaFX and Java. I also have plans to encorperate JBox2D and modify the world to allow Agents to interact with other Agents using collision. Just so you know I do anticipate that the performance impact will be too much and I will need implement Box2D in c++ but its worth a try in my opinion. Look for the project by the end of the month.
If anyone has a good idea of how I should gage performance of this application please reach out to me.
Julian Hershey
Mar 30, 2016, 2:42:33 PM3/30/16
to Scriptbots
Hello Skid Runner! I saw your fork on Github before this message. It has been a long-thought-about project to convert to Java to take advantage of its cross-platform nature. I'm a little too ingrained in the C++ version however, so I wish you the best of luck, and look forward to updates if you make them!
Skid Runner
Jun 30, 2016, 10:45:50 PM6/30/16
to Scriptbots
Can someone explain the brain names for me?
Julian Hershey
Jul 7, 2016, 12:57:14 AM7/7/16
to Scriptbots
I'll try my best. I didn't design the brains
Assembly was Andrej's latest experimental version. It was intended to replace the MLPBrain. It includes some experimental operators, but is unfinished, and thus far I haven't seen a need for it. I think it was supposed to be faster than the MLP
DWRAON, stands for Damped, Weighted, Recurrent, AND-OR Network. It was the one we started with, and is described in his video of Version 4 above. Each box has several CONNectionS, and chose either to be an AND type or an OR type, thus multiplying or adding the different, weighted connection values together. It had some limitations, such as flooring or exploding connections, making the bots behave "jitterly".
MLPBrain, stands for Multi-Layer Perception. Not sure where the name comes from, but it's basically the DWRAON but without Or connections, and every brain box is sent through a mathematical function called a sigmoid, which essentially allows us to feed every brain box very large positive or negative values to get a wider range of responses. It's very natural (no brain connection is ever truely "off" or "on" [0,1]) and produces more responses to changes in the environment. It is the current brain used by both me and Andrej's latest versions.
Hope that answers your question!
Assembly was Andrej's latest experimental version. It was intended to replace the MLPBrain. It includes some experimental operators, but is unfinished, and thus far I haven't seen a need for it. I think it was supposed to be faster than the MLP
DWRAON, stands for Damped, Weighted, Recurrent, AND-OR Network. It was the one we started with, and is described in his video of Version 4 above. Each box has several CONNectionS, and chose either to be an AND type or an OR type, thus multiplying or adding the different, weighted connection values together. It had some limitations, such as flooring or exploding connections, making the bots behave "jitterly".
MLPBrain, stands for Multi-Layer Perception. Not sure where the name comes from, but it's basically the DWRAON but without Or connections, and every brain box is sent through a mathematical function called a sigmoid, which essentially allows us to feed every brain box very large positive or negative values to get a wider range of responses. It's very natural (no brain connection is ever truely "off" or "on" [0,1]) and produces more responses to changes in the environment. It is the current brain used by both me and Andrej's latest versions.
Hope that answers your question!
Message has been deleted
Skid Runner
Jul 25, 2016, 5:24:27 PM7/25/16
to Scriptbots
Thank you this helps a lot.
Assembly == SOM
DWRAON == RNN
MLPBrain == MLP
Reply all
Reply to author
Forward
0 new messages