Current ToDos

• finish agent code:

  • set up NN, play around with the layers, the ones from the tutorial might not be the best X
  • handle caching, function to sample cache X
  • how to handle checkpoints X -> PPO fallback implementation

• features:

  • implement wayfinding (a_star) X
  • moving kernels? (window_method) X
  • cluster coins? (idw_map) X
  • how to handle players with diffrent amount of points (maybe target players with higher points?)
  • killing an opponent
  • less features is a better idea (better idea) X -> calculating feature importance (random forest) --> droping a feature from a certain of predictors --> some paper or resource