In [1]:

    

import tensorflow as tf


    

In [2]:

    

tf.zeros([3, 3])


    

    
Out[2]:





<tf.Tensor 'zeros:0' shape=(3, 3) dtype=float32>

In [3]:

    

tf.ones([3, 3])


    

    
Out[3]:





<tf.Tensor 'ones:0' shape=(3, 3) dtype=float32>

In [4]:

    

tf.fill([3, 3], 42)


    

    
Out[4]:





<tf.Tensor 'Fill:0' shape=(3, 3) dtype=int32>

In [5]:

    

constant_tsr = tf.constant([1, 2, 3])
constant_tsr


    

    
Out[5]:





<tf.Tensor 'Const:0' shape=(3,) dtype=int32>

In [6]:

    

tf.zeros_like(constant_tsr)


    

    
Out[6]:





<tf.Tensor 'zeros_like:0' shape=(3,) dtype=int32>

In [7]:

    

tf.ones_like(constant_tsr)


    

    
Out[7]:





<tf.Tensor 'ones_like:0' shape=(3,) dtype=int32>

In [8]:

    

tf.linspace(0.0, 1.0, 3)


    

    
Out[8]:





<tf.Tensor 'LinSpace:0' shape=(3,) dtype=float32>

In [9]:

    

tf.range(6, 15, 3)


    

    
Out[9]:





<tf.Tensor 'range:0' shape=(3,) dtype=int32>

In [10]:

    

tf.random_uniform([3, 3], minval=0, maxval=1)


    

    
Out[10]:





<tf.Tensor 'random_uniform:0' shape=(3, 3) dtype=float32>

In [11]:

    

tf.random_normal([3, 3], mean=0.0, stddev=1.0)


    

    
Out[11]:





<tf.Tensor 'random_normal:0' shape=(3, 3) dtype=float32>

In [12]:

    

tf.truncated_normal([3, 3], mean=0.0, stddev=1.0)


    

    
Out[12]:





<tf.Tensor 'truncated_normal:0' shape=(3, 3) dtype=float32>

In [13]:

    

my_var = tf.Variable(tf.zeros([2, 3]))
sess = tf.Session()
initialize_op = tf.global_variables_initializer()
sess.run(initialize_op)


    

In [30]:

    

import numpy as np
sess = tf.Session()
x = tf.placeholder(tf.float32, shape=[2, 2])
y = tf.identity(x)
x_vals = np.random.rand(2, 2)
sess.run(y, feed_dict={x: x_vals})


    

    
Out[30]:





array([[ 0.84918463,  0.20622016],
       [ 0.73926389,  0.94779098]], dtype=float32)

In [31]:

    

import numpy as np
import tensorflow as tf
sess = tf.Session()


    

In [38]:

    

identity_matrix = tf.diag([1.0, 1.0, 1.0])
print(identity_matrix)
print(sess.run(identity_matrix))


    

    




Tensor("Diag_2:0", shape=(3, 3), dtype=float32)
[[ 1.  0.  0.]
 [ 0.  1.  0.]
 [ 0.  0.  1.]]

In [36]:

    

A = tf.truncated_normal([2, 3])
B = tf.fill([2, 3], 5.0)
C = tf.random_uniform([3, 2])
D = tf.convert_to_tensor(np.array(
    [[1., 2., 3.],
     [-3., -7., -1.],
     [0., 5., -2.]]
))


    

In [46]:

    

print(sess.run(B))


    

    




[[ 5.  5.  5.]
 [ 5.  5.  5.]]

In [43]:

    

print(sess.run(A + B))


    

    




[[ 5.19373512  4.94957685  3.75595045]
 [ 4.53172874  4.03519726  3.87743998]]

In [44]:

    

print(sess.run(B - B))


    

    




[[ 0.  0.  0.]
 [ 0.  0.  0.]]

In [45]:

    

print(sess.run(tf.matmul(B, identity_matrix)))


    

    




[[ 5.  5.  5.]
 [ 5.  5.  5.]]

In [47]:

    

sess.run(C)


    

    
Out[47]:





array([[ 0.80826008,  0.66252875],
       [ 0.14581406,  0.29976416],
       [ 0.08765745,  0.3801676 ]], dtype=float32)

In [48]:

    

sess.run(tf.transpose(C))


    

    
Out[48]:





array([[ 0.57005787,  0.04247582,  0.94901741],
       [ 0.64883411,  0.5056982 ,  0.37611508]], dtype=float32)

In [49]:

    

sess.run(tf.matrix_determinant(D))


    

    
Out[49]:





-38.0

In [50]:

    

sess.run(tf.matrix_inverse(D))


    

    
Out[50]:





array([[-0.5       , -0.5       , -0.5       ],
       [ 0.15789474,  0.05263158,  0.21052632],
       [ 0.39473684,  0.13157895,  0.02631579]])

In [53]:

    

aaa = tf.matmul(D, tf.matrix_inverse(D))
aaa


    

    
Out[53]:





<tf.Tensor 'MatMul_2:0' shape=(3, 3) dtype=float64>

In [54]:

    

sess.run(aaa)


    

    
Out[54]:





array([[  1.00000000e+00,   0.00000000e+00,  -8.32667268e-17],
       [ -1.11022302e-16,   1.00000000e+00,   1.04083409e-16],
       [  1.11022302e-16,   0.00000000e+00,   1.00000000e+00]])

In [55]:

    

sess.run(tf.cholesky(identity_matrix))


    

    
Out[55]:





array([[ 1.,  0.,  0.],
       [ 0.,  1.,  0.],
       [ 0.,  0.,  1.]], dtype=float32)

In [56]:

    

sess.run(tf.self_adjoint_eig(D))


    

    
Out[56]:





(array([-10.65907521,  -0.22750691,   2.88658212]),
 array([[ 0.21749542,  0.63250104, -0.74339638],
        [ 0.84526515,  0.2587998 ,  0.46749277],
        [-0.4880805 ,  0.73004459,  0.47834331]]))

In [57]:

    

sess.run(tf.div(3, 4))


    

    
Out[57]:





0

In [58]:

    

sess.run(tf.truediv(3, 4))


    

    
Out[58]:





0.75

In [59]:

    

sess.run(tf.div(3.0, 4.0))


    

    
Out[59]:





0.75

In [60]:

    

sess.run(tf.floordiv(3.0, 4.0))


    

    
Out[60]:





0.0

In [61]:

    

sess.run(tf.mod(22.0, 5.0))


    

    
Out[61]:





2.0

In [62]:

    

def custom_polynomial(value):
    return(tf.subtract(3 * tf.square(value), value) + 10)


    

In [67]:

    

sess.run(custom_polynomial(11))


    

    
Out[67]:





362

In [75]:

    

sess.run(tf.nn.relu([-3., 3., 10.]))


    

    
Out[75]:





array([  0.,   3.,  10.], dtype=float32)

In [76]:

    

sess.run(tf.nn.relu6([-3., 3., 10.]))


    

    
Out[76]:





array([ 0.,  3.,  6.], dtype=float32)

In [77]:

    

sess.run(tf.nn.sigmoid([-1., 0., 1.]))


    

    
Out[77]:





array([ 0.26894143,  0.5       ,  0.7310586 ], dtype=float32)

In [78]:

    

sess.run(tf.nn.tanh([-1., 0., 1.]))


    

    
Out[78]:





array([-0.76159418,  0.        ,  0.76159418], dtype=float32)

In [79]:

    

sess.run(tf.nn.softsign([-1., 0., 1.]))


    

    
Out[79]:





array([-0.5,  0. ,  0.5], dtype=float32)

In [91]:

    

import matplotlib.pyplot as plt
%matplotlib inline
xlist = np.linspace(-5, 5, 1000)
ss = sess.run(tf.nn.softsign(xlist))
sp = sess.run(tf.nn.softplus(xlist))
el = sess.run(tf.nn.elu(xlist))
plt.plot(xlist, ss)
plt.plot(xlist, sp)
plt.plot(xlist, el)


    

    
Out[91]:





[<matplotlib.lines.Line2D at 0x113acee80>]






    

In [87]:

    

sess.run(tf.nn.softplus([-1., 0., 1.]))


    

    
Out[87]:





array([ 0.31326166,  0.69314718,  1.31326163], dtype=float32)

In [93]:

    

from sklearn import datasets
iris = datasets.load_iris()


    

In [95]:

    

len(iris.data)


    

    
Out[95]:





150

In [96]:

    

len(iris.target)


    

    
Out[96]:





150

In [98]:

    

iris.data[0]


    

    
Out[98]:





array([ 5.1,  3.5,  1.4,  0.2])

In [99]:

    

iris.target[0]


    

    
Out[99]:





0

In [100]:

    

from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets('MNIST_data/', one_hot=True)


    

    




Extracting MNIST_data/train-images-idx3-ubyte.gz
Extracting MNIST_data/train-labels-idx1-ubyte.gz
Extracting MNIST_data/t10k-images-idx3-ubyte.gz
Extracting MNIST_data/t10k-labels-idx1-ubyte.gz

In [101]:

    

len(mnist.train.images)


    

    
Out[101]:





55000

In [102]:

    

len(mnist.test.images)


    

    
Out[102]:





10000

In [103]:

    

len(mnist.validation.images)


    

    
Out[103]:





5000