Ai::Neat

This gem is base on NEAT-JS. That have the main purpose to convert the code from javascript to ruby.

Installation

Add this line to your application's Gemfile:

gem 'ai-neat'

And then execute:

$ bundle install

Or install it yourself as:

$ gem install ai-neat

Usage

1. Get started with config

config = {
  models: [
    { node_count: 5, node_type: :input },
    { node_count: 3, node_type: :output, activationfunc: :softmax }
  ],
  mutation_rate: 0.1,
  crossover_method: :random,
  mutation_method: :random,
  population_size: population_size
}

models parameter are array of layers of NEAT at least their are needed 2 layers are node_type = :input and :output but it's possible to add any :middle layer between of them.

  models: [
    { node_count: 5, node_type: :input },
    { node_count: 5, node_type: :inpu, activationfunc: :softmax },
    { node_count: 3, node_type: :output, activationfunc: :softmax }
  ]

And activationfunc are needed if that layer is not output layer. Current supported activationfunc are :relu, :tanh, :sigmoid, :leaky_relu, :softmax.

Supported crossover_method are :random and :slice.

Supported mutation_method are :random and :edit.

1. Initiate Neat Class from previous config

neat = Ai::Neat::Neat.new(config)

1. Set generation loop

2. Set inputs array equal to number of node_type = :input to each Creature [i]

neat.set_inputs(inputs, i)

If your simmulation have various creature life time like a gameover you can skip to set inputs that creature.

1. When set inputs compleated

neat.feed_forward

1. Get decisions for each creature in array

decisions = neat.decisions

returned decisions value are array of integer that started from 0 to node_count - 1 of :output layer.

1. Finish generation

neat.do_gen

And you can get best creature result by

neat.best_creature

This function will return index of best creature in lastest generation.

Example

population_size = 100

config = {
models: [
    { node_count: 5, node_type: :input },
    { node_count: 3, node_type: :output, activationfunc: :softmax }
],
mutation_rate: 0.1,
crossover_method: :random,
mutation_method: :random,
population_size: population_size
}

neat = Ai::Neat::Neat.new(config)

scores = population_size.times.map { 0 }
first_score = 0

# Generation
100.times.each do |gen|
scores = population_size.times.map { 0 }

# play
50.times.each do
    inputs = 5.times.map { rand(-1.0..1.0) }

    (0..(scores.count - 1)).each do |i|
        neat.set_inputs(inputs, i)
    end

    neat.feed_forward
    decisions = neat.decisions

    (0..(scores.count - 1)).each do |i|
        if inputs.last > 0
            case decisions[i]
            when 0
            scores[i] += 1
            when 1
            scores[i] -= 1
            end
        else
            case decisions[i]
            when 0
            scores[i] -= 1
            when 1
            scores[i] += 1
            end
        end

        scores[i] = 0 if scores[i] < 0
    end
end

(0..(scores.count - 1)).each do |i|
    neat.set_fitness(scores[i], i)
end

first_score = scores[neat.best_creature] if gen == 0

neat.do_gen
end

exported = neat.export

neat2 = Ai::Neat::Neat.new(config)
neat2.import(exported)

Development

After checking out the repo, run bin/setup to install dependencies. Then, run rake spec to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and the created tag, and push the .gem file to rubygems.org.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/Sunama/ai-neat. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the code of conduct.

License

The gem is available as open source under the terms of the MIT License.

Code of Conduct

Everyone interacting in the Ai::Neat project's codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.