brains 0.1.0-java → 0.1.1-java

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 0d640546a10c8984b876f7f95feeab5cddc10f39
-  data.tar.gz: ed2c32d65a0cd4cda1ca00667f821b95e5cc70c7
+  metadata.gz: 555149c0570a03a1ef8af988f8114013457cc2ca
+  data.tar.gz: 01fb0792ebd5b148ea5ab0da4f43cf5da4d482a4
 SHA512:
-  metadata.gz: 95bb48840ee28b6d10f527b0b2efc876db9da382669e5e66027305bd92651846ec9343a217dba1fef4a9d95366cb240d1aeb08a8c73aa98062e58785b81291ec
-  data.tar.gz: 77f3df827145cd2314238a5eb0c09f9f52bbf016dd3a6cc9786c472fe5298f7d2e014edd3642bdc1155f15c0bf630176b6317ac45779d16e5cbf52a07c03b222
+  metadata.gz: 96f733e7575451619ae2c354921059cb4c2b06524d69d14433b0aeb3b202054caf79cf6ea42153e69e14d16a8c3e551de004629b4189d24e648e39a92c49807f
+  data.tar.gz: ee83b4c978701cfd4095cd0d8de754e8e371fe0e543648d8f1cb4c625c265cd52d9ab3de4b3a14212f5b3f27ea0bfc6d15b530ce10a55d0b1651c37f43a4de39

data/README.md

@@ -2,7 +2,9 @@
 
 A Feedforward neural network toolkit for JRuby. Easily add machine learning features
 to your ruby application using this Gem. Though there are faster native C implementations
-this java backend provides a balance of performance and ease of use.
+available (e.g. FANN) we need something that is simple, beginner friendly and just works.
+
+This java based implementation provides a balance of performance and ease of use.
 
 ## Installation
 
@@ -23,6 +25,12 @@ Or install it yourself as:
 
     $ gem install brains
 
+## Features
+
+* Customizable network parameters depending on requirements
+* Fast (A bit slower than FANN but significantly faster than a pure ruby implementation)
+* NN backend implementation in Java which allows for platform agnostic implementation
+
 ## Usage
 
 The brains gem contains facilities for training and using the feedforward neural network
@@ -38,9 +46,8 @@ require 'brains'
 
 # Build a 3 layer network: 4 input neurons, 4 hidden neurons, 3 output neurons
 # Bias neurons are automatically added to input + hidden layers; no need to specify these
-# 5 = 4 in one hidden layer + 1 output neuron (input neurons not counted)
 
-nn = Brains::Net.create(2, 1, 5, { neurons_per_layer: 4 })
+nn = Brains::Net.create(2 /* no. of inputs */, 1 /*no. of outputs */, 1 /*hidden layer*/, { neurons_per_layer: 4 })
 nn.randomize_weights
 ```
 
@@ -133,6 +140,10 @@ You can then save it to a file. You can then load it back using load()
 
 ```ruby
   nn = Brains::Net.load(saved_state)
+
+
+  # use
+  nn.feed([0.9, 0.9])
 ```
 
 For other samples please take a look at the example folder.

data/example/colors.rb

@@ -0,0 +1,45 @@
+#!/usr/bin/env ruby
+
+require 'brains'
+
+#This neural network will identify the main color name based on rgb values
+
+label_encodings = {
+  "Red"   => [1, 0, 0],
+  "Green" => [0, 1, 0],
+  "Blue"  => [0, 0 ,1]
+}
+#0000ff
+training_data = [
+  [[0xf0/0xff.to_f, 0xf8.to_f/0xff.to_f, 1],[0,0,1]],
+  [[0x00/0xff.to_f, 0x00.to_f/0xff.to_f, 1],[0,0,1]],
+  [[0x00/0xff.to_f, 0x00.to_f/0xff.to_f, 0x8b.to_f/0xff.to_f],[0,0,1]]
+]
+
+nn = Brains::Net.create(2, 1, 9, { neurons_per_layer: 4 })
+nn.randomize_weights
+
+
+# test on untrained data
+#0000ee
+test_data = [
+  [0x00/0xff.to_f, 0x00.to_f/0xff.to_f, 0xee.to_f/0xff.to_f],
+]
+
+results = test_data.collect { |item|
+  nn.feed(item)
+}
+
+p results
+
+result = nn.optimize(training_data, 0.01, 1_000 ) { |i, error|
+  puts "#{i} #{error}"
+}
+
+puts "after training"
+
+results = test_data.collect { |item|
+  nn.feed(item)
+}
+
+p results

data/example/iris.rb

@@ -50,7 +50,7 @@ end
 
 # Build a 3 layer network: 4 input neurons, 4 hidden neurons, 3 output neurons
 # Bias neurons are automatically added to input + hidden layers; no need to specify these
-nn = Brains::Net.create(4, 3, 7, { neurons_per_layer: 4 })
+nn = Brains::Net.create(4, 3, 1, { neurons_per_layer: 4 })
 nn.randomize_weights
 
 prediction_success = -> (actual, ideal) {

data/example/xor.rb

@@ -7,7 +7,7 @@ require 'brains'
 # Bias neurons are automatically added to input + hidden layers; no need to specify these
 # 5 = 4 in one hidden layer + 1 output neuron (input neurons not counted)
 
-nn = Brains::Net.create(2, 1, 5, { neurons_per_layer: 4 })
+nn = Brains::Net.create(2, 1, 1, { neurons_per_layer: 4 })
 nn.randomize_weights
 
 # A    B   A XOR B
@@ -37,7 +37,7 @@ results = test_data.collect { |item|
 
 p results
 
-result = nn.optimize(training_data, 0.01, 1_000 ) { |i, error|
+result = nn.optimize(training_data, 0.01, 1_000_000 ) { |i, error|
   puts "#{i} #{error}"
 }
 

data/lib/brains/net.rb

@@ -4,13 +4,16 @@ module Brains
   class Net
     attr_accessor :nn, :config
 
-    def self.create(input, output, total, opts = {})
-      config = com.dayosoft.nn.NeuralNet::Config.new(input, output, total)
+    def self.create(input, output, total_hidden_layers = 1, opts = {})
+      neurons_per_layer = opts[:neurons_per_layer] || 5
+
+      config = com.dayosoft.nn.NeuralNet::Config.new(input, output,  total_hidden_layers * neurons_per_layer + output)
       config.bias = opts[:bias] || 1.0
       config.outputBias = opts[:output_bias] || 1.0
       config.learningRate = opts[:learning_rate] || 0.1
-      config.neuronsPerLayer = opts[:neurons_per_layer] || 5
+      config.neuronsPerLayer = neurons_per_layer
       config.momentumFactor = opts[:momentum_factor] || 0.5
+      config.backPropagationAlgorithm = opt_t_back_alg(opts[:train_method] || :standard)
       config.activationFunctionType = opt_to_func(opts[:activation_function] || :htan)
       config.outputActivationFunctionType = opt_to_func(opts[:activation_function] || :sigmoid)
       config.errorFormula = opt_to_error_func(opts[:activation_function] || :mean_squared)
@@ -38,7 +41,7 @@ module Brains
       @nn.dumpWeights.to_a.map(&:to_a)
     end
 
-    def optimize(test_cases, target_error = 0.01, max_epoch = 1_000_000_000, is_batch = false, &callback)
+    def optimize(test_cases, target_error = 0.01, max_epoch = 1_000_000_000, callback_interval = 1000, &callback)
       inputs = []
       outputs = []
 
@@ -47,7 +50,8 @@ module Brains
         outputs << item[1].to_java(Java::double)
       end
 
-      result = @nn.optimize(java.util.ArrayList.new(inputs), java.util.ArrayList.new(outputs), target_error, max_epoch, is_batch, callback)
+      result = @nn.optimize(java.util.ArrayList.new(inputs), java.util.ArrayList.new(outputs), target_error, max_epoch,
+        callback_interval, callback)
       { iterations: result.first, error: result.second }
     end
 
@@ -91,6 +95,15 @@ module Brains
       end
     end
 
+    def self.opt_t_back_alg(func)
+      case func
+      when :standard
+        com.dayosoft.nn.NeuralNet::Config::STANDARD_BACKPROPAGATION
+      when :rprop
+        com.dayosoft.nn.NeuralNet::Config::RPROP_BACKPROPAGATION
+      end
+    end
+
     def self.opt_to_error_func(func)
       case func
       when :mean_squared

data/lib/brains/version.rb

@@ -1,3 +1,3 @@
 module Brains
-  VERSION = "0.1.0"
+  VERSION = "0.1.1"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: brains
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.1.1
 platform: java
 authors:
 - Joseph Emmanuel Dayo
@@ -11,47 +11,47 @@ cert_chain: []
 date: 2017-02-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
+  name: bundler
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1.12'
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '1.12'
-  name: bundler
   prerelease: false
   type: :development
+- !ruby/object:Gem::Dependency
+  name: rake
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.12'
-- !ruby/object:Gem::Dependency
+        version: '10.0'
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '10.0'
-  name: rake
   prerelease: false
   type: :development
+- !ruby/object:Gem::Dependency
+  name: rspec
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '10.0'
-- !ruby/object:Gem::Dependency
+        version: '3.0'
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '3.0'
-  name: rspec
   prerelease: false
   type: :development
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: '3.0'
 description: A feedforward neural network library for JRuby
 email:
 - joseph.dayo@gmail.com
@@ -70,6 +70,7 @@ files:
 - bin/console
 - bin/setup
 - brains.gemspec
+- example/colors.rb
 - example/iris.rb
 - example/xor.rb
 - iris.data