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
|
#!/usr/bin/env ruby
require 'zipf'
require 'matrix'
require 'optimist'
def read_data fn
f = ReadFile.new fn
data = []
while line = f.gets
line.strip!
a = []
a << 1.0
tokenize(line).each { |i| a << i.to_f }
v = Vector.elements a
data << v
end
return data
end
def dot x, y
r = 0.0
x.each_with_index { |_,j|
r += x[j] * y[j]
}
return r
end
def approx_eql x, y, eps=10**-10
return false if !x||!y
return false if x.size!=y.size
x.each_with_index { |_,i|
return false if (x[i]-y[i]).abs>eps
}
return true
end
def main
conf = Optimist::options do
opt :input, "input data", :type => :string, :required => true
opt :output, "1/0 output data", :type => :string, :required => true
end
data = read_data conf[:input]
dim = data[0].size
zeros = [0.0]*dim
t = ReadFile.readlines(conf[:output]).map{ |i| i.to_f }
model = Vector.elements zeros
prev_model = nil
gradient = Vector.elements zeros
hessian = Matrix.build(dim,dim) { |i,j| 0.0 }
i = 0
while true
i += 1
data.each_with_index { |x,j|
m = 1.0/(1+Math.exp(-dot(model, x)))
gradient += (m-t[j]) * x
hup = Matrix.column_vector(x) * Matrix.row_vector(x)
hessian += m*(1.0-m) * hup
}
gradient /= data.size
hessian /= data.size
model -= hessian.inverse * gradient
break if approx_eql model, prev_model
prev_model = model
end
STDERR.write "ran for #{i} iterations\n"
puts model.to_s
end
main
|
