brainz

Brainz

Artificial Neural Network Library written in Ruby.

Supported training algorithms

• Backpropagation

Installation

Add this line to your application's Gemfile:

gem 'brainz'

And then execute:

$ bundle

Or install it yourself as:

$ gem install brainz

Usage

Simple example, logical AND operation

require 'brainz'

brainz = Brainz::Brainz.new

brainz.teach do |iteration, error|
  that(1, 1).is(1)
  that(1, 0).is(0)
  that(0, 1).is(0)
  that(0, 0).is(0)
  p "error_rate = #{'%.3f' % error || 0 } after #{iteration} iterations" if iteration % 10 == 0
end

puts "0 and 0 = #{brainz.guess(a: 0, b: 0)}"
puts "0 and 1 = #{brainz.guess(a: 0, b: 1)}"
puts "1 and 1 = #{brainz.guess(a: 1, b: 1)}"
puts "1 and 0 = #{brainz.guess(a: 1, b: 0)}"

Advanced usage, ligical XOR operation

require 'brainz'

brainz = Brainz::Brainz.new

# specify number of hidden layers
# 3 hidden layers with 4, 7 and 3 neurons
brainz.num_hidden = [4, 7, 3]

# tune learning parameters: learning_rate, momentum, wanted_error (mse)
brainz.teach(learning_rate: 0.2, momentum: 0.05, wanted_error: 0.01) do |iteration, error|
  that(1, 1).is(0)
  that(1, 0).is(1)
  that(0, 1).is(1)
  that(0, 0).is(0)
end

puts "Learning took #{brainz.last_iterations}, error: #{brainz.error}, time: #{brainz.learning_time} s."

Contributing

• Fork it
• Create your feature branch (git checkout -b my-new-feature)
• Commit your changes (git commit -am 'Added some feature')
• Push to the branch (git push origin my-new-feature)
• Create new Pull Request