optimuscrime / ntnu-2016-it3708-project3-4 Goto Github PK

• Change namespace org.thomas.project3 -> org.thomas.annea
• Abstract settings file
• Rename Main
• Make it possible to use None as activation function
• Minimal activation limit
• Graphs

• Main file with various stuff
• Fix GUI things
• Extend network
• Add support for bias
• Add support for gains
• Own config files for both projects

The report should be max 2 pages

• Document Genotype/Phenotype and the conversion between
• Describe how you have implemented the CTRNN, how it is used and how it determines actions
• Describe runs for all 3 scenarios (standard, pull and no-wrap)
• Analyse an evolved CTRNN
  • Show the evolved weights
  • Analyse how these weights determine the behaviour of the agent
  • Try different inputs to the network and analyze the outputs. Try to show how the memory of the nework can give different ouput for the samen input based on eralier input

Getting random number of food and poison on boards, ranging for 17-30 ish.
I have still used this for creating plot for the report, but we might need to redo this.

There should be exaclty 33.3% food on the map and 22.2% poison on the map.

The report should be max 4 pages

• Configuration
• Fitness Function
• ANN
• EA Performance
• Compare performance (very small, 2-3 lines tops)