Refactor the neural net benchmark

time/bench_neural_net.rb

@@ -1,131 +1,136 @@
 # from bryanbibat's gist: https://gist.github.com/2348802
 
 class Synapse
+  WEIGHT_RANGE = (-1.0..1.0).freeze
+
   attr_accessor :weight, :prev_weight
-  attr_accessor :source_neuron, :dest_neuron
+  attr_reader :source_neuron, :dest_neuron
 
-  def initialize(source_neuron, dest_neuron, prng)
-    self.source_neuron = source_neuron
-    self.dest_neuron = dest_neuron
-    self.prev_weight = self.weight = prng.rand(-1.0..1.0)
+  def initialize(source_neuron:, dest_neuron:)
+    @source_neuron = source_neuron
+    @dest_neuron = dest_neuron
+    @prev_weight = @weight = rand(WEIGHT_RANGE)
   end
 end
 
 class Neuron
-
   LEARNING_RATE = 1.0
   MOMENTUM = 0.3
+  WEIGHT_RANGE = (-1.0..1.0).freeze
 
-  attr_accessor :synapses_in, :synapses_out
-  attr_accessor :threshold, :prev_threshold, :error
+  attr_reader :synapses_in, :synapses_out
+  attr_reader :threshold, :prev_threshold, :error
   attr_accessor :output
 
-  def initialize(prng)
-    self.prev_threshold = self.threshold = prng.rand(-1.0..1.0)
-    self.synapses_in = []
-    self.synapses_out = []
+  def initialize
+    @synapses_in = []
+    @synapses_out = []
+    @prev_threshold = @threshold = rand(WEIGHT_RANGE)
   end
 
   def calculate_output
-    # calculate output based on the previous layer
-    # use logistic function
-
-    activation = synapses_in.inject(0.0) do |sum, synapse| 
-      sum + synapse.weight * synapse.source_neuron.output 
-    end
-    activation -= threshold
+    activation = synapses_in.sum do |synapse|
+      synapse.weight * synapse.source_neuron.output
+    end - @threshold
 
-    self.output = 1.0 / (1.0 + Math.exp(-activation))
-  end
-
-  def derivative
-    output * (1 - output)
+    @output = 1.fdiv(Math.exp(-activation) + 1)
   end
 
-  def output_train(rate, target)
-    self.error = (target - output) * derivative
+  def output_train(rate:, target:)
+    @error = (target - @output) * derivative
     update_weights(rate)
   end
 
-  def hidden_train(rate)
-    self.error = synapses_out.inject(0.0) do |sum, synapse| 
-      sum + synapse.prev_weight * synapse.dest_neuron.error
+  def hidden_train(rate:)
+    @error = synapses_out.sum do |synapse|
+      synapse.prev_weight * synapse.dest_neuron.error
     end * derivative
     update_weights(rate)
   end
 
+  private
+
+  def derivative
+    @output * (1 - @output)
+  end
+
   def update_weights(rate)
-    synapses_in.each do |synapse|
-      temp_weight = synapse.weight
-      synapse.weight += (rate * LEARNING_RATE * error * synapse.source_neuron.output) +
-        (MOMENTUM * ( synapse.weight - synapse.prev_weight))
-      synapse.prev_weight = temp_weight
+    update_synapses_in(rate)
+    update_thresholds(rate)
+  end
+
+  def update_synapses_in(rate)
+    @synapses_in.each do |synapse|
+      prev_weight = synapse.weight
+      synapse.weight += LEARNING_RATE * rate * @error * synapse.source_neuron.output
+      synapse.weight += MOMENTUM * (synapse.weight - synapse.prev_weight)
+      synapse.prev_weight = prev_weight
     end
-    temp_threshold = threshold
-    self.threshold += (rate * LEARNING_RATE * error * -1) + 
-      (MOMENTUM * (threshold - prev_threshold))
-    self.prev_threshold = temp_threshold
+  end
+
+  def update_thresholds(rate)
+    prev_threshold = @threshold
+    @threshold += LEARNING_RATE * rate * @error * -1
+    @threshold += MOMENTUM * (@threshold - @prev_threshold)
+    @prev_threshold = prev_threshold
   end
 end
 
 class NeuralNetwork
-  attr_accessor :prng
-
-  def initialize(inputs, hidden, outputs)
-    self.prng = Random.new
+  def initialize(inputs:, hidden:, outputs:)
+    @input_layer  = Array.new(inputs) { Neuron.new }
+    @hidden_layer = Array.new(hidden) { Neuron.new }
+    @output_layer = Array.new(outputs) { Neuron.new }
 
-    @input_layer = (1..inputs).map { Neuron.new(prng) }
-    @hidden_layer = (1..hidden).map { Neuron.new(prng) }
-    @output_layer = (1..outputs).map { Neuron.new(prng) }
-
-    @input_layer.product(@hidden_layer).each do |source, dest|
-      synapse = Synapse.new(source, dest, prng)
-      source.synapses_out << synapse
-      dest.synapses_in << synapse
-    end
-    @hidden_layer.product(@output_layer).each do |source, dest|
-      synapse = Synapse.new(source, dest, prng)
-      source.synapses_out << synapse
-      dest.synapses_in << synapse
-    end
+    add_synapses(@input_layer, @hidden_layer)
+    add_synapses(@hidden_layer, @output_layer)
   end
 
-  def train(inputs, targets)
+  def train(inputs:, targets:)
     feed_forward(inputs)
 
     @output_layer.zip(targets).each do |neuron, target|
-      neuron.output_train(0.3, target)
+      neuron.output_train(rate: 0.3, target: target)
     end
-    @hidden_layer.each { |neuron| neuron.hidden_train(0.3) }
+    @hidden_layer.each { |neuron| neuron.hidden_train(rate: 0.3) }
   end
 
   def feed_forward(inputs)
-    @input_layer.zip(inputs).each do |neuron, input|
-      neuron.output = input
-    end
-    @hidden_layer.each { |neuron| neuron.calculate_output }
-    @output_layer.each { |neuron| neuron.calculate_output }
+    @input_layer.zip(inputs) { |neuron, input| neuron.output = input }
+    @hidden_layer.each(&:calculate_output)
+    @output_layer.each(&:calculate_output)
   end
 
   def current_outputs
-    @output_layer.map { |neuron| neuron.output }
+    @output_layer.map(&:output)
   end
 
+  private
+
+  def add_synapses(source_layer, dest_layer)
+    source_layer.product(dest_layer) do |source_neuron, dest_neuron|
+      synapse = Synapse.new(source_neuron: source_neuron,
+                            dest_neuron: dest_neuron)
+
+      source_neuron.synapses_out << synapse
+      dest_neuron.synapses_in << synapse
+    end
+  end
 end
 
 require 'benchmark'
 
-(ARGV[0] || 5).to_i.times do
-  x = Benchmark.measure do |x|
-    xor = NeuralNetwork.new(2, 10, 1)
-  
-    10000.times do 
-      xor.train([0, 0], [0])
-      xor.train([1, 0], [1])
-      xor.train([0, 1], [1])
-      xor.train([1, 1], [0])
+ARGV.fetch(0) { 5 }.to_i.times do
+  results = Benchmark.measure do
+    xor = NeuralNetwork.new(inputs: 2, hidden: 10, outputs: 1)
+
+    10_000.times do
+      xor.train(inputs: [0, 0], targets: [0])
+      xor.train(inputs: [1, 0], targets: [1])
+      xor.train(inputs: [0, 1], targets: [1])
+      xor.train(inputs: [1, 1], targets: [0])
     end
-  
+
     xor.feed_forward([0, 0])
     puts xor.current_outputs
     xor.feed_forward([0, 1])
@@ -135,5 +140,7 @@ def current_outputs
     xor.feed_forward([1, 1])
     puts xor.current_outputs
   end
-  puts x
+
+  puts results
 end
+