/
main.rb
170 lines (148 loc) · 4.77 KB
/
main.rb
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
$LOAD_PATH.unshift File.expand_path('../../../lib', __FILE__)
require 'open3'
require 'sabina'
DIM = 2
K = 3
EPOCH = 100
xrange = [-2.2, 2.2]
yrange = [-2.2, 2.2]
training_data = Sabina::MultilayerPerceptron.load_csv('training_data.csv')
# You can create your own layer class.
# In this example, a rectified linear function is set
# as an activation function.
# @f_ is differentiation of @f.
class MyHiddenLayer < Sabina::Layer::BaseLayer
def initialize(size)
super
@f = ->(x){ x > 0.0 ? x : 0.0 }
@f_ = ->(x){ x > 0.0 ? 1.0 : 0.0 }
end
end
options = {
:layers => [
Sabina::Layer::MPInputLayer.new(DIM),
MyHiddenLayer.new(16),
MyHiddenLayer.new(8),
Sabina::Layer::MPOutputLayer.new(K)
],
:mini_batch_size => 10,
:learning_rate => 0.01,
:training_data => training_data,
}
mp = Sabina::MultilayerPerceptron.new(options)
##################################################
# Make a chart of trainig data.
##################################################
Open3.popen3('gnuplot') do |gp_in, gp_out, gp_err|
output_file = "./mp_training_data.png"
gp_in.puts "set colorsequence classic"
gp_in.puts "set terminal png size 480, 450"
gp_in.puts "set output '#{output_file}'"
gp_in.puts "set label 2 center at screen 0.5,0.9 'Training Data' font 'Helvetica,22'"
gp_in.puts "set tmargin 3.2"
gp_in.puts "set xlabel 'x0'"
gp_in.puts "set ylabel 'x1'"
gp_in.puts "set size square"
gp_in.puts xrange.tap { |f, t| break "set xrange [#{f}:#{t}]" }
gp_in.puts yrange.tap { |f, t| break "set yrange [#{f}:#{t}]" }
plot = "plot "
K.times do |k|
plot << "'-' notitle pt 1 ps 1 lc #{k+1},\\\n"
end
plot.gsub!(/,\\\n\z/, "\n")
K.times do |k|
label = Array.new(K) { |j| j == k ? 1 : 0 }
training_data.each do |data|
if label == data[:d]
data[:x].tap { |x, y| plot << "#{x}, #{y}\n" }
end
end
plot << "e\n"
end
gp_in.puts plot
puts output_file
gp_in.puts "set output"
gp_in.puts "exit"
gp_in.close
end
##################################################
# Make a GIF of the learning process.
##################################################
Open3.popen3('gnuplot') do |gp_in, gp_out, gp_err|
output_file = "./mp_learning_process.gif"
gp_in.puts "set colorsequence classic"
gp_in.puts "set terminal gif animate delay 10 optimize size 880, 450"
gp_in.puts "set output '#{output_file}'"
gp_in.puts "set tmargin 3.2"
gp_in.puts "set size square"
log = []
tmp_data = xrange.map { |v| (20*v).to_i }.tap { |f, t| break [*f..t] }
.product yrange.map { |v| (20*v).to_i }.tap { |f, t| break [*f..t] }
tmp_data.map! do |data|
data.each_with_object(0.05).map(&:*)
end
x_mat = Matrix.columns( tmp_data )
EPOCH.times do |t|
mp.learn
log << mp.error(training_data)
gp_in.puts "set multiplot layout 1, 2"
##################################################
# Training Data
##################################################
gp_in.puts "set label 2 center at screen 0.28,0.9 'Training Data' font 'Helvetica,22'"
gp_in.puts xrange.tap { |f, t| break "set xrange [#{f}:#{t}]" }
gp_in.puts yrange.tap { |f, t| break "set yrange [#{f}:#{t}]" }
gp_in.puts "set xlabel 'x0'"
gp_in.puts "set ylabel 'x1'"
gp_in.puts "unset grid"
plot = "plot "
K.times do |k|
plot << "'-' notitle pt 1 ps 1 lc #{k+1},\\\n"
end
K.times do |k|
plot << "'-' notitle pt 7 ps 1 lc #{k+1},\\\n"
end
plot.gsub!(/,\\\n\z/, "\n")
y_ary = mp.propagate_forward(x_mat).t.to_a
K.times do |k|
tmp_data.each_with_index do |(x, y), n|
label = y_ary[n].index( y_ary[n].max )
plot << "#{x}, #{y}\n" if label == k
end
plot << "e\n"
end
K.times do |k|
label = Array.new(K) { |j| j == k ? 1 : 0 }
training_data.each do |data|
if label == data[:d]
data[:x].tap { |x, y| plot << "#{x}, #{y}\n" }
end
end
plot << "e\n"
end
gp_in.puts plot
##################################################
# Training Error
##################################################
gp_in.puts "set label 2 center at screen 0.79,0.9 'Training Error' font 'Helvetica,22'"
gp_in.puts "set xrange [0:#{EPOCH}]"
gp_in.puts "set yrange [0:#{log.first + 10}]"
gp_in.puts "set xlabel 'iteration number'"
gp_in.puts "set ylabel 'training error'"
gp_in.puts "set grid"
plot = "plot "
plot << "'-' notitle with lines lw 3 lt 1 lc 1,\\\n"
plot.gsub!(/,\\\n\z/, "\n")
log.each { |x, y| plot << "#{x}, #{y}\n" }
plot << "e\n"
gp_in.puts plot
# progress bar
puts " error : #{log.last}"
puts " [#{("*"*((t.to_f / EPOCH)*10).to_i).ljust(9, " ")}]"
print "\e[2A"; STDOUT.flush;
end
gp_in.puts "set output"
puts output_file
gp_in.puts "exit"
gp_in.close
end