Many copies of a single reinforcement learning agent trained to play snake using the deepq learning algorithm in convnetjs. The food is hidden but the agents all see the same food item, leading to a mixture of random and coordinated movements across the screen. The agents were trained on a smaller 5x5 grid, which is what they continue to see, repeated across the canvas.

https://github.com/Henry-E/snake-visualisation

More information

Published319 days ago
StatusIn development
PlatformsHTML5
Rating
(1)
Authorhenry-e
GenreOther
Tagsscreensaverjam
Player countSingleplayer

Comments

Log in with your itch.io account to leave a comment.

That's AWESOME!