Hacker News new | past | comments | ask | show | jobs | submit login
Evolving Neural Networks for a Flatland Agent (veierland.net)
4 points by pveierland on April 23, 2016 | hide | past | favorite | 1 comment



If you wait until the simulation is complete, a buton thay says "random" appear just below the map and you can see how one of the pacmans (the champion?) performs in the map.

I tried to change "foodEaten - 2 * poisonEaten" to "foodEaten - 10 * poisonEaten" but then the pacman walks a little and then it starts to go in a "circle"(square) up>left>down>right>[repeat]. It get's too cautious :(.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: