Flethuseo Flethuseo - 6 months ago 43
Ruby Question

Training neural network for XOR in Ruby

I am trying to train a feedforward network to work to perform an XOR operations with the Ruby Library AI4R. However,
when I evaluate for the XOR after training it. I am not getting the correct output. Has anyone used this library before
and gotten it to learn the XOR operation.

I am using two input neurons, three neurons in a hidden layer, and one layer for the output, as I saw a precomputed
XOR feed forward neural network like this before.

require "rubygems"
require "ai4r"

# Create the network with:
# 2 inputs
# 1 hidden layer with 3 neurons
# 1 outputs
net = Ai4r::NeuralNetwork::Backpropagation.new([2, 3, 1])

example = [[0,0],[0,1],[1,0],[1,1]]
result = [[0],[1],[1],[0]]

# Train the network
400.times do |i|
j = i % result.length
puts net.train(example[j], result[j])
end

# Use it: Evaluate data with the trained network
puts "evaluate 0,0: #{net.eval([0,0])}" # => evaluate 0,0: 0.507531383375123
puts "evaluate 0,1: #{net.eval([0,1])}" # => evaluate 0,1: 0.491957823618629
puts "evaluate 1,0: #{net.eval([1,0])}" # => evaluate 1,0: 0.516413912471401
puts "evaluate 1,1: #{net.eval([1,1])}" # => evaluate 1,1: 0.500197884691668


Ted

Answer

You haven't trained it for enough iterations. If you change 400.times to 8000.times you'll come much closer (and closer still at 20000.times).

At 20000.times, I get

puts "evaluate 0,0: #{net.eval([0,0])}"  # =>  evaluate 0,0: 0.030879848321403
puts "evaluate 0,1: #{net.eval([0,1])}"  # =>  evaluate 0,1: 0.97105714994505
puts "evaluate 1,0: #{net.eval([1,0])}"  # =>  evaluate 1,0: 0.965055940880282
puts "evaluate 1,1: #{net.eval([1,1])}"  # =>  evaluate 1,1: 0.0268317078331645

You can also increase net.learning_rate (but not too much).

Comments