In [1]:

    

from physlearn.NeuralNet.NeuralNet import NeuralNet
import numpy
import tensorflow as tf
from math import pi
import matplotlib.pyplot as plt
%matplotlib inline


    

In [2]:

    

x_train = numpy.linspace(0, 2 * pi, 30).reshape(1, 30)
y_train = numpy.sin(x_train[0]).reshape(1, 30)


    

In [3]:

    

x_cv = numpy.linspace(0, 2 * pi, 1000).reshape(1, 1000)
y_cv = numpy.sin(x_cv[0]).reshape(1, 1000)


    

In [4]:

    

plt.plot(x_train[0], y_train[0], 'x', color='red')


    

    
Out[4]:





[<matplotlib.lines.Line2D at 0x7fadb7495160>]






    

In [5]:

    

net = NeuralNet(-1, 1)


    

In [6]:

    

net.add_input_layer(1)
net.add(10, tf.sigmoid)
net.add_output_layer(1)


    

In [7]:

    

net.compile()


    

In [8]:

    

cost_list = net.train('prediction', x_train, y_train, 5, 3000, 0.1)
plt.plot(list(map(lambda item: item ** (-1), cost_list)))


    

    




100%|██████████| 3000/3000 [00:01<00:00, 1959.27it/s]






    
Out[8]:





[<matplotlib.lines.Line2D at 0x7fadb3c03320>]






    

In [9]:

    

net.calculate_cost(x_train, y_train)


    

    
Out[9]:





0.06241226434886385

In [10]:

    

y_pred = net.run(x_cv)
plt.plot(x_train[0], y_train[0], 'x', color='red')
plt.plot(x_cv[0], y_pred[0])


    

    
Out[10]:





[<matplotlib.lines.Line2D at 0x7fadb1a3c518>]






    

In [11]:

    

res = net.unroll_matrixes()


    

In [12]:

    

res[0]


    

    
Out[12]:





[(array([[ 0.86298793],
         [ 1.56153218],
         [ 1.46124137],
         [ 1.26388663],
         [-0.79724256],
         [ 2.05805715],
         [ 0.70013601],
         [ 0.86209423],
         [ 1.23698465],
         [-0.26954899]]), array([[ 0.45230029],
         [ 0.0773675 ],
         [ 0.06125173],
         [-2.54524229],
         [-0.76043917],
         [-1.9324716 ],
         [ 0.20241035],
         [-0.22628199],
         [-2.38871301],
         [-0.09356024]])),
 (array([[-0.8177541 ,  1.35590382,  0.87282782, -2.21011461,  0.6381469 ,
           2.70699917, -0.17095081, -0.36841195, -1.7560022 , -0.84169342]]),
  array([[-0.16331892]]))]

In [13]:

    

unroll_vector = res[2]


    

In [14]:

    

net.roll_matrixes(unroll_vector)


    

    
Out[14]:





[(array([[ 0.86298793],
         [ 1.56153218],
         [ 1.46124137],
         [ 1.26388663],
         [-0.79724256],
         [ 2.05805715],
         [ 0.70013601],
         [ 0.86209423],
         [ 1.23698465],
         [-0.26954899]]), array([[ 0.45230029],
         [ 0.0773675 ],
         [ 0.06125173],
         [-2.54524229],
         [-0.76043917],
         [-1.9324716 ],
         [ 0.20241035],
         [-0.22628199],
         [-2.38871301],
         [-0.09356024]])),
 (array([[-0.8177541 ,  1.35590382,  0.87282782, -2.21011461,  0.6381469 ,
           2.70699917, -0.17095081, -0.36841195, -1.7560022 , -0.84169342]]),
  array([[-0.16331892]]))]