the_noggin 0.0.1

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA1:
+  metadata.gz: f2c52db539548af3d06e7ce7e291d44439f4610b
+  data.tar.gz: a03adc11a237bec24c7dd0ae8a59cd59ee9ec4b3
+SHA512:
+  metadata.gz: 3608bb2ae571b06ad72c9ed67387c3c436f79b42e4933ef5bf5e9d988e5681afeefb4650c23c964b653faf975774f7e5a14b12caed4baad135be48f55f1cbaf7
+  data.tar.gz: 5eb96f0af2ae976f8fe1fb35a50fc6134f8e30afd2e92375e07eaaab637db1afebc78df51c9b380a22d0f9bb773ff2637a6b809e86e1b2290b53fa3033adaa78

data/.gitignore

@@ -0,0 +1,18 @@
+*.gem
+*.rbc
+.bundle
+.config
+.yardoc
+Gemfile.lock
+InstalledFiles
+_yardoc
+coverage
+doc/
+lib/bundler/man
+pkg
+rdoc
+spec/reports
+test/tmp
+test/version_tmp
+tmp
+.idea

data/Gemfile

@@ -0,0 +1,3 @@
+source 'https://rubygems.org'
+
+gemspec

data/LICENSE.txt

@@ -0,0 +1,22 @@
+Copyright (c) 2015 Shawn
+
+MIT License
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice shall be
+included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
+LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

data/README.md

@@ -0,0 +1,47 @@
+# Noggin
+Ruby Neural Network implementation using backpropagation and gradient descent for training
+
+
+``` Ruby
+network = Noggin::Network.new
+
+network.train([
+  { input: [0, 0], output: 0 },
+  { input: [0, 1], output: 1 },
+  { input: [1, 0], output: 1 },
+  { input: [1, 1], output: 0 }
+])
+
+network.run [0, 0]  # 0.0163
+network.run [0, 1]  # 0.9573
+network.run [1, 0]  # 0.9702
+network.run [1, 1]  # 0.0142
+
+```
+
+## Options
+``` Ruby
+Noggin::Network.new( 
+    max_training_laps: 100000, # How many propgation of errors to do when training
+    learning_rate: 0.1, # How fast the network learns
+    hidden_layer_size: 1 , # Number of hidden layers
+    hidden_layer_node_size: 2 # Number of nodes each hidden layer has
+)
+```
+
+## Print Network
+
+``` Ruby
+network.pretty_print
+```
+```
+ ------                                           ------                                          --------------      
+|      | -EDGE--(w: 0.438443, d: 0.01759)        |      | -EDGE--(w: 0.515923, d: 0.09704)       | ed: 0.668486       
+|      | -EDGE--(w: 0.746539, d: 0.013825)        ------                                         | d: 0.148145        
+ ------                                           ------                                         | e: 0.223437        
+ ------                                          |      | -EDGE--(w: 0.485781, d: 0.11099)       | o: 0.668486        
+|      | -EDGE--(w: 0.199745, d: 0.01759)         ------                                          --------------      
+|      | -EDGE--(w: 0.345684, d: 0.013825)       
+ ------                                          
+   
+```

data/Rakefile

@@ -0,0 +1 @@
+require "bundler/gem_tasks"

data/lib/noggin/network.rb

@@ -0,0 +1,104 @@
+module Noggin
+  class Network
+
+    attr_reader :input_nodes
+    attr_reader :output_node
+    attr_reader :layers
+    attr_reader :options
+
+    DEFAULTS = {
+        learning_rate: 0.2,
+        max_training_laps: 100000,
+        hidden_layer_size: 1,
+        hidden_layer_node_size: 2
+    }
+
+    def initialize **opts
+      @options = DEFAULTS.merge opts
+      @ready = false
+      @layers = []
+    end
+
+    def run input
+      update_input_nodes input
+      output_node.output
+    end
+
+    def train data_batch
+      init_network(data_batch)unless @ready
+      options[:max_training_laps].times do
+        data_batch.each do |batch|
+          propagate_error! batch[:input], batch[:output]
+        end
+      end
+      return self
+    end
+
+    def propagate_error! input, expected
+      update_input_nodes input
+      output_node.expected = expected
+      input_nodes.each { |node| node.derivative_chain }
+      update_weights!
+    end
+
+    def update_weights!
+      edges.each do |edge|
+        edge.weight -= options[:learning_rate] * edge.derivative
+      end
+    end
+
+    def init_network data_batch
+      @input_nodes = Array.new(data_batch.first[:input].size){ Noggin::Node::Input.new }
+      @layers << @input_nodes
+      last_layer = @input_nodes
+      options[:hidden_layer_size].times do |i|
+        new_layer = Array.new(options[:hidden_layer_node_size]){ Noggin::Node::Base.new }
+        @layers << new_layer
+        connect_layer last_layer, new_layer
+        last_layer = new_layer
+      end
+      @output_node = Noggin::Node::Output.new
+      @layers << [@output_node]
+      last_layer.each { |node| connect_nodes(node, output_node) }
+      @ready = true
+    end
+
+    def edges
+      edges = []
+      queue = [output_node]
+      while queue.size != 0 do
+        node = queue.pop
+        node.origins.each do |edge|
+          edges << edge
+          queue << edge.origin
+        end
+      end
+      edges
+    end
+
+    def connect_nodes origin, dest
+      edge = Noggin::Node::Edge.new origin: origin, dest: dest
+      origin.dests << edge
+      dest.origins << edge
+    end
+
+    def connect_layer origins, dests
+      origins.each do |origin|
+        dests.each do |dest|
+          connect_nodes origin, dest
+        end
+      end
+    end
+
+    def update_input_nodes input
+      input_nodes.each_with_index do | node, i |
+        node.output = input[i]
+      end
+    end
+
+    def pretty_print
+       Noggin::PrettyPrinter.print_network layers
+    end
+
+  end
+end

data/lib/noggin/node/base.rb

@@ -0,0 +1,42 @@
+module Noggin
+  module Node
+    class Base
+
+      attr_reader :origins
+      attr_reader :dests
+      attr_accessor :derivative
+
+      def initialize
+        @origins = []
+        @dests = []
+      end
+
+      def input
+        origins.inject(0) { |sum, edge | sum += edge.value }
+      end
+
+      def output
+        1 / ( 1 + Math.exp(-1 * input) )
+      end
+
+      def output_derivative
+        output * (1 - output)
+      end
+
+      def derivative_chain
+        derivative = output_derivative * dests.inject(0) { |sum, edge| sum += edge.derivative_chain }
+      end
+
+      def pretty_print
+        out = []
+        out << " ------"
+        dests.each do |edge|
+          out << "|      | -EDGE--(#{edge.pretty_print}) "
+        end
+        out << " ------"
+        out
+      end
+
+    end
+  end
+end

data/lib/noggin/node/edge.rb

@@ -0,0 +1,35 @@
+module Noggin
+  module Node
+    class Edge
+
+      attr_accessor :origin
+      attr_accessor :dest
+      attr_accessor :weight
+      attr_accessor :derivative
+
+      def initialize origin: origin, dest: dest, weight: rand(0.20...0.80)
+        @origin = origin
+        @dest = dest
+        @weight = weight
+      end
+
+      def input
+        origin.output
+      end
+
+      def value
+        origin.output * weight
+      end
+
+      def derivative_chain
+        @derivative = input * dest.derivative_chain
+        @weight * dest.derivative_chain
+      end
+
+      def pretty_print
+        "w: #{@weight.round(6)}, d: #{@derivative.round(6)}"
+      end
+
+    end
+  end
+end

data/lib/noggin/node/input.rb

@@ -0,0 +1,7 @@
+module Noggin
+  module Node
+    class Input < Noggin::Node::Base
+      attr_accessor :output
+    end
+  end
+end

data/lib/noggin/node/output.rb

@@ -0,0 +1,31 @@
+module Noggin
+  module Node
+    class Output < Noggin::Node::Base
+
+      attr_accessor :expected
+
+      def error
+        0.5 * (expected - output)**2
+      end
+
+      def error_derivative
+        output - expected
+      end
+
+      def derivative_chain
+        output_derivative * error_derivative
+      end
+
+      def pretty_print
+        out = []
+        out << " --------------"
+        out << "| ed: #{error_derivative.round(6)}"
+        out << "| d: #{derivative_chain.round(6)}"
+        out << "| e: #{error.round(6)}"
+        out << "| o: #{output.round(6)}"
+        out << " --------------"
+        out
+      end
+    end
+  end
+end

data/lib/noggin/pretty_printer.rb

@@ -0,0 +1,26 @@
+module Noggin
+  class PrettyPrinter
+    def self.print_network layers
+      grid = []
+      layers.each do |layer|
+        grid << col = []
+        layer.each do |node|
+          col << node.pretty_print
+        end
+        col.flatten!
+      end
+      grid[0].zip(*grid[1..-1]).each do |row|
+        row.each_with_index do |cell, col_i|
+          max_length = grid[col_i].max_by{|s| s.size }.size
+          if cell
+            room = max_length - cell.length
+            print cell
+            print " " * room
+            print "      "
+          end
+        end
+        print "\n"
+      end
+    end
+  end
+end

data/lib/noggin/version.rb

@@ -0,0 +1,3 @@
+module Noggin
+  VERSION = '0.0.1'
+end

data/lib/noggin.rb

@@ -0,0 +1,6 @@
+require 'noggin/node/base'
+require 'noggin/node/input'
+require 'noggin/node/output'
+require 'noggin/node/edge'
+require 'noggin/network'
+require 'noggin/pretty_printer'

data/spec/network_spec.rb

@@ -0,0 +1,37 @@
+require_relative '../lib/noggin'
+
+describe Noggin::Network do 
+
+  subject { Noggin::Network.new( max_training_laps: 1, learning_rate: 0.1, hidden_layer_size: 1, hidden_layer_node_size: 1 ) }
+
+  let(:input_node) { subject.layers[0].first }
+  let(:hidden_node) { subject.layers[1].first }
+  let(:output_node) { subject.layers[2].first }
+    
+  before do
+    subject.train [{ input: [1], output: 0 }]
+    
+  end
+
+  it 'sets up the network graph according to settings' do
+    expect(input_node.dests.first.dest).to eq hidden_node
+    expect(hidden_node.dests.first.dest).to eq output_node
+  end
+
+  it 'sets hidden layer size' do
+    expect(subject.layers.size).to eq(3)
+  end
+
+  it 'sets hidden layer node size' do
+    expect(subject.layers[1].size).to eq(1)
+  end
+
+  it 'sets max max training laps' do
+  
+  end
+
+  it 'backpropagates error' do
+  
+  end
+
+end

metadata

@@ -0,0 +1,104 @@
+--- !ruby/object:Gem::Specification
+name: the_noggin
+version: !ruby/object:Gem::Version
+  version: 0.0.1
+platform: ruby
+authors:
+- Shawn
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2015-02-21 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: bundler
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1.3'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1.3'
+- !ruby/object:Gem::Dependency
+  name: rake
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '10.4'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '10.4'
+- !ruby/object:Gem::Dependency
+  name: rspec
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '3.2'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '3.2'
+description: Ruby Neural Network implementation using backpropagation and gradient
+  descent for training
+email:
+- shaw3257@gmail.com
+executables: []
+extensions: []
+extra_rdoc_files: []
+files:
+- ".gitignore"
+- Gemfile
+- LICENSE.txt
+- README.md
+- Rakefile
+- lib/noggin.rb
+- lib/noggin/network.rb
+- lib/noggin/node/base.rb
+- lib/noggin/node/edge.rb
+- lib/noggin/node/input.rb
+- lib/noggin/node/output.rb
+- lib/noggin/pretty_printer.rb
+- lib/noggin/version.rb
+- spec/network_spec.rb
+homepage: https://github.com/shaw3257/noggin
+licenses:
+- MIT
+metadata: {}
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubyforge_project: 
+rubygems_version: 2.4.3
+signing_key: 
+specification_version: 4
+summary: Pass in training samples, and the network will try to find pathways that
+  lead to the least amount of error. The network is customizable in that it let’s
+  you control the learning rate, hidden layer size and depth, and max training iterations.
+test_files:
+- spec/network_spec.rb