In [1]:

    

import tensorflow as tf
import numpy as np


    

In [2]:

    

xy = np.loadtxt('../data/softmax_data.txt',unpack=True, dtype='float32')
x_data = np.transpose(xy[0:3])
y_data = np.transpose(xy[3:])


    

In [3]:

    

np.transpose(xy)


    

    
Out[3]:





array([[ 1.,  2.,  1.,  0.,  0.,  1.],
       [ 1.,  3.,  2.,  0.,  0.,  1.],
       [ 1.,  3.,  4.,  0.,  0.,  1.],
       [ 1.,  5.,  5.,  0.,  1.,  0.],
       [ 1.,  7.,  5.,  0.,  1.,  0.],
       [ 1.,  2.,  5.,  0.,  1.,  0.],
       [ 1.,  6.,  6.,  1.,  0.,  0.],
       [ 1.,  7.,  7.,  1.,  0.,  0.]], dtype=float32)

In [4]:

    

X = tf.placeholder("float",[None, 3])
Y = tf.placeholder("float",[None, 3])

W = tf.Variable(tf.zeros([3,3]))
h = tf.nn.softmax(tf.matmul(X,W))
learning_rate = 0.001

cost = tf.reduce_mean(-tf.reduce_sum(Y*tf.log(h),reduction_indices=1))
opt = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost)
init = tf.global_variables_initializer()

sess = tf.Session()
sess.run(init)
for step in range(2000):
    sess.run(opt, feed_dict={X:x_data, Y:y_data})
    if step % 200 == 0:
        print(step, sess.run(cost,feed_dict={X:x_data,Y:y_data}),sess.run(W))


    

    




0 1.09774 [[ -8.33333252e-05   4.16666626e-05   4.16666480e-05]
 [  1.66666665e-04   2.91666714e-04  -4.58333379e-04]
 [  1.66666723e-04   4.16666706e-04  -5.83333313e-04]]
200 1.05962 [[-0.02051385 -0.00103983  0.02155367]
 [ 0.01406439  0.01097753 -0.0250419 ]
 [ 0.01431208  0.03574874 -0.05006079]]
400 1.04985 [[-0.04282598 -0.00625899  0.04908496]
 [ 0.01747187  0.00156368 -0.01903554]
 [ 0.01831204  0.04954105 -0.06785304]]
600 1.0407 [[-0.0651786  -0.01176361  0.07694222]
 [ 0.01943522 -0.00848972 -0.01094547]
 [ 0.0211558   0.06118308 -0.0823388 ]]
800 1.03194 [[-0.08734014 -0.01729389  0.10463405]
 [ 0.0211172  -0.01796172 -0.00315547]
 [ 0.02396628  0.07198346 -0.09594975]]
1000 1.02354 [[-0.10928575 -0.02282182  0.13210757]
 [ 0.02266254 -0.02678035  0.00411784]
 [ 0.02685852  0.08210357 -0.10896213]]
1200 1.01547 [[-0.13101467 -0.02834092  0.15935561]
 [ 0.02409703 -0.0349762   0.01087924]
 [ 0.02983199  0.09159798 -0.12143002]]
1400 1.0077 [[-0.15252906 -0.03384715  0.18637618]
 [ 0.02543302 -0.04258837  0.01715543]
 [ 0.03287465  0.10050803 -0.13338272]]
1600 1.00021 [[-0.1738314  -0.03933692  0.21316831]
 [ 0.02668082 -0.04965463  0.02297391]
 [ 0.03597424  0.10887133 -0.14484566]]
1800 0.992979 [[-0.19492455 -0.04480688  0.23973146]
 [ 0.02784993 -0.05621056  0.02836075]
 [ 0.03911952  0.11672311 -0.15584266]]

In [5]:

    

a = sess.run(h, feed_dict={X:[[1,11,7]]})
print(a,sess.run(tf.arg_max(a,1)))

b = sess.run(h, feed_dict={X:[[1,3,4]]})
print(b,sess.run(tf.arg_max(b,1)))

c = sess.run(h, feed_dict={X:[[1,1,0]]})
print(c,sess.run(tf.arg_max(c,1)))

all = sess.run(h, feed_dict={X:[[1,1,0],[1,3,4],[1,1,0]]})
print(all,sess.run(tf.arg_max(all,1)))


    

    




[[ 0.46269614  0.35486573  0.18243809]] [0]
[[ 0.33821005  0.42100322  0.24078675]] [1]
[[ 0.27005655  0.29087913  0.43906435]] [2]
[[ 0.27005655  0.29087913  0.43906435]
 [ 0.33821005  0.42100322  0.24078675]
 [ 0.27005655  0.29087913  0.43906435]] [2 1 2]