Snake with NEAT: benchmarking changes to an evolutionary algorithm
Run-to-run noise, a board small enough to measure on, and three ways I fooled myself
Read the postJim Wright · software engineer
Notes from building software, mostly in Go.
29 posts across 2 series: evolving neural networks with NEAT, teaching a snake to play itself, and what I learn along the way.
Latest
Before that
All 29 posts- Snake with NEAT: a hunger input to break stalemates A good network does not die any more. It gets stuck - and the fix is one number that has nothing to do with the board Snake AI in Go: eleven ways to play 8 min
- Snake with NEAT: training runs and checkpoints Population, generations, checkpoints, and why the fittest genome is not the one you keep Snake AI in Go: eleven ways to play 8 min
- Snake with NEAT: designing the fitness function Three terms, in descending order of how much they are allowed to matter, and the trap in each of them Snake AI in Go: eleven ways to play 6 min
- Snake with NEAT: designing the network's inputs Eighteen numbers, why each one is there, and the reading bug that was worth more than any of them Snake AI in Go: eleven ways to play 8 min
- Snake in Go: evolving a neural network with NEAT Nobody tells this one how to play. It is a neural network that was evolved rather than written Snake AI in Go: eleven ways to play 5 min
- Snake in Go: building a Hamiltonian cycle with search Growing a route into a Hamiltonian cycle, and the two small decisions that took it from 65 to winning every game Snake AI in Go: eleven ways to play 7 min
- Snake in Go: safe shortcuts along the tour The same tour, jumping ahead on it whenever that can be shown to be safe - and a third fewer moves Snake AI in Go: eleven ways to play 5 min
Series
Longer runs of posts that build on each other. Start at part one.
- 16 parts Snake AI in Go: eleven ways to play One game, eleven players, and what each of them is worth. From a snake moving at random, through the searches and the tours that never lose, to a neural network that was evolved rather than …
- 12 parts NEAT in Go: neuroevolution from scratch Building neatgo a piece at a time: the network, the genome, historical markings, mutation, crossover, speciation, selection and reproduction, and the run that ties them together.
Who writes this
I'm Jim, a software engineer. I write here to solidify my understanding of a topic, and to keep track of what I've been working on.
About me