t_learn 0.1.1.2 → 0.1.1.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: bcf9731528ca98ed43cfe5f2a7686707def2e6a9
-  data.tar.gz: a90853e05203740a90c3c5ce6cf9fb13f0ab36bf
+  metadata.gz: 4997f379b3ef4527598a83ceb94ab6620d61b886
+  data.tar.gz: cca4c6643f596a70d69b1a79a86ca94189bcce0d
 SHA512:
-  metadata.gz: 7b25db6487317f85009cb795bec1060e71cdca91804ce74f013e8933d3dfc6f1128c65d02829cac5956b31747cc159907f92fecdb7a910ef18219c95759aa8a5
-  data.tar.gz: 86dfa12d25b29010c0e6ccebbb120e2f28062e70bdece7e43d830e79d503038fd808553218f0a8d621f11f800be22cd2e30b8e940c5c752b7ba28e008e60a403
+  metadata.gz: 1951ea181fb690965373c0444af8c3d3682b2bdf17889baf7a4aa6d84f57701a5a4211916a1034f83aef801b6101ba64a1da1a5c2f43fa67f3b605dc8c807dd7
+  data.tar.gz: fc1ddcaa6a905309de2e8729fd96f44c11ac04dc8c652287d8c5e805957af42af3206c627276767557b4e1f6ddd1889ae4568834fcb0023d87743942033cbbd4

data/README.md

@@ -25,15 +25,17 @@ respect for keras.
 
 ``` ruby
 
-model = TLearn::FeedForwardNeuralNetwork.new
+require "t_learn"
+
+model = TLearn::FeedForwardNeuralNetwork.new(learning_rate=0.1)
 
-model.add_layer(2)
-model.add_layer(3)
-model.add_layer(1)
+model.add_layer(node_num=2)
+model.add_layer(node_num=3)
+model.add_layer(node_num=1)
 
 x_train = [[0.0, 0.0],[0.0, 1.0], [1.0, 0.0], [1.0, 1.0]]
 y_train = [[ 0.0 ], [ 1.0 ],[ 1.0 ],[ 0.0 ]]
-model.fit(x_train, y_train, 50000)
+model.fit(x_train, y_train, epoch=50000)
 
 x_test = x_train
 y_test = y_train
@@ -42,6 +44,10 @@ err_rate = model.evaluate(x_test, y_test)
 
 puts "err rate: #{err_rate}%"
 
+# p x_test[0]
+# model.propagation(x_text[0])
+# puts model.get_output_layer[1].w
+
 ```
 
 ### result

data/lib/t_learn/feedforward_neural_network.rb

@@ -2,18 +2,19 @@
 # -*- encoding: utf-8 -*-
 
 module TLearn
-
   #
   # ==
   #
   class FeedForwardNeuralNetwork
     attr_accessor :layer_list, :layer_size, :link_list, :node_id, :learning_rate, :err_list, :threshold
-    def initialize(learning_rate=0.1, threshold=0.0)
+    def initialize(learning_rate=0.1, threshold=0.0, momentum_rate=0.01)
       @layer_size = 0    #layer iterator
       @layer_list = Array.new
       @link_list = Hash.new
       @node_id = 0
       @learning_rate = learning_rate
+      @momentum_weight_list = Hash.new
+      @momentum_rate = momentum_rate
       @err_list = Array.new
       @threshold = threshold
     end
@@ -43,6 +44,7 @@ module TLearn
       @layer_list[@layer_size - 1].each do |from_node|
         @layer_list[@layer_size].each do |to_node|
           @link_list["#{from_node.id}_#{to_node.id}"] = rand(-1.0...1.0)
+          @momentum_weight_list["#{from_node.id}_#{to_node.id}"] = 0.0
         end
       end
     end
@@ -69,6 +71,7 @@ module TLearn
     end
 
     def propagation(x)
+      raise "input size is different from  node num of input layer "  if @layer_list[0].size != x.size
       # input data 
       @layer_list[0].each_with_index do |node, i|
         node.input (x[i])
@@ -85,6 +88,8 @@ module TLearn
           end
         end
       end
+
+      return get_output_layer 
     end
 
     def calc_ave_err(y)
@@ -96,25 +101,31 @@ module TLearn
       return ave_err 
     end
 
-    
+
     #
     # === 
     # 
     # @param y Array teacher_data
     #
     def back_propagation(y)
+
+      raise "output size different from node num of output layer"  if get_output_layer.size != y.size
+      # raise "o"  if get_output_layer.size != y.size
       delta = {}
       ( @layer_size - 1).downto(0) do |layer_num|
         if ( @layer_size - 1) == layer_num   # if output layer
           @layer_list[layer_num].each_with_index do |output_node, i|
-              delta["#{output_node.id}"] = -1.0 * calc_err(y[i], output_node.w) * output_node.w * (1.0 -output_node.w)
+            delta["#{output_node.id}"] = -1.0 * calc_err(y[i], output_node.w) * output_node.w * (1.0 -output_node.w)
           end
         else 
           @layer_list[layer_num].each do |from_node|
             # リンクの更新
             @layer_list[layer_num + 1].each do |to_node|
+              # モメンタムによる更新
+              momentum_weight = @momentum_rate * @momentum_weight_list["#{from_node.id}_#{to_node.id}"]
               update_weight = -1.0 * @learning_rate * delta["#{to_node.id}"] * from_node.w
-              @link_list["#{from_node.id}_#{to_node.id}"] = @link_list["#{from_node.id}_#{to_node.id}"] + update_weight 
+              @link_list["#{from_node.id}_#{to_node.id}"] = @link_list["#{from_node.id}_#{to_node.id}"] + update_weight + momentum_weight
+              @momentum_weight_list["#{from_node.id}_#{to_node.id}"] = update_weight # for momentum
             end
             # その層のdeltaの更新
             delta["#{from_node.id}"] = calc_delta(delta,layer_num, from_node) * from_node.w * (1.0 - from_node.w)
@@ -147,14 +158,18 @@ module TLearn
           err += (y_f - o.w).abs
         end
         sum_err += (err/y_test[0].size)
-        puts "x #{x}, y #{y} , output #{output}"
+        # puts "x #{x}, y #{y} , output #{output}"
       end 
       return (sum_err/y_test.size) * 100.0
       # return 0.0
     end
 
     def get_output_layer
-      return @layer_list[@layer_size-1]
+      output = []
+      @layer_list[@layer_size-1].each do |node|
+        output.push(node.w)
+      end
+      return output
     end
 
     class Node

data/lib/t_learn/k_means.rb

@@ -0,0 +1,71 @@
+#!/usr/bin/ruby
+# -*- encoding: utf-8 -*-
+
+require "yaml"
+
+module TLearn
+  class K_Means
+    attr_accessor 
+    
+    def initialize
+          end
+ 
+    #
+    # === make datas for test
+    #
+    def make_data
+
+    end
+
+    def run
+      
+    end
+
+    
+ 
+    def calc_burden_ratio(k)
+         end    
+
+    def log_likelihood
+       
+    end 
+
+    def update_mu(k, burden_ratio)
+          end
+
+    def update_var(k, burden_ratio)
+          end
+
+    def update_pi(k)
+    end
+    
+    #
+    # === 混合正規分布を生成する
+    #
+    def gaussian_mix()
+          end
+
+
+    #
+    # === gauusian distribution 
+    #
+    def gauusian(x, mu, sigma)
+      f1 = (1.0/Math.sqrt(2.0*Math::PI))
+      f2 = Math.exp(-((x-mu)^2 / 2*sigma^2))
+      return f1 * f2
+    end
+
+    #
+    # ===ボックス―ミューラー法をよる正規分布乱数発生
+    # @param mu flout 平均
+    # @param sigma flout 標準偏差
+    # @return ボックスミューラー法に従う正規分布に従う乱数を生成
+    #
+    def normal_rand(mu = 0,sigma = 1.0)
+      a, b = rand(), rand() ;
+      return (Math.sqrt(-2*Math.log(rand()))*Math.sin(2*Math::PI*rand()) * sigma) + mu
+    end
+
+
+  end
+end

data/lib/t_learn/version.rb

@@ -1,3 +1,3 @@
 module TLearn
-  VERSION = "0.1.1.2"
+  VERSION = "0.1.1.3"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: t_learn
 version: !ruby/object:Gem::Version
-  version: 0.1.1.2
+  version: 0.1.1.3
 platform: ruby
 authors:
 - Tcom242242
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2017-01-01 00:00:00.000000000 Z
+date: 2017-05-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -60,7 +60,6 @@ extensions: []
 extra_rdoc_files: []
 files:
 - ".gitignore"
-- ".rake_tasks~"
 - ".rspec"
 - ".travis.yml"
 - CODE_OF_CONDUCT.md
@@ -74,6 +73,7 @@ files:
 - lib/t_learn/em.rb
 - lib/t_learn/feedforward_neural_network.rb
 - lib/t_learn/hop_field_net.rb
+- lib/t_learn/k_means.rb
 - lib/t_learn/version.rb
 - t_learn.gemspec
 homepage: https://github.com/Tcom242242/t_learn

data/.rake_tasks~

@@ -1,7 +0,0 @@
-build
-clean
-clobber
-install
-install:local
-release[remote]
-spec