Skip to content

A novel tensor encoding method used with the genetic algorithm that plays Snake at superhuman levels.

Notifications You must be signed in to change notification settings

willjhliang/encoded-neuroevolution

Repository files navigation

Mastering Snake With Neuroevolution Search and Tensor Decomposition Optimization

A collection of algorithms, models, encoders, and tests exploring the use of encoding methods in a genetic algorithm that trains neural networks to play Snake. Finalized optimal model used a modified genetic algorithm on a small CNN encoded with tensor decomposition. The network successfully plays Snake at superhuman levels, reaching a score of over 30 before ending the game.

This work is included in Tensor Shape Search for Optimum Data Compression by Solgi et al.

About

A novel tensor encoding method used with the genetic algorithm that plays Snake at superhuman levels.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages