In [1]:

    

from physlearn.NeuralNet.NeuralNetPro import NeuralNetPro
from physlearn.Optimizer import RandomEvolutionStrategy
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 0x7f73ce653438>]






    

In [5]:

    

net = NeuralNetPro(-1, 1)


    

In [6]:

    

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


    

In [7]:

    

net.compile()


    

In [8]:

    

net.set_train_type('prediction')


    

In [9]:

    

dim = net.return_unroll_dim()


    

In [10]:

    

net.set_random_matrixes()


    

In [11]:

    

def cost(params):
    net.roll_matrixes(params)
    return net.calculate_cost(x_train, y_train)


    

In [12]:

    

res = RandomEvolutionStrategy.optimize(cost, dim, 0.35, 10, 100000, 0.001)
cost_list = res[0]
plt.plot(cost_list)


    

    




100%|██████████| 10000/10000 [01:04<00:00, 154.34it/s]






    
Out[12]:





[<matplotlib.lines.Line2D at 0x7f73c8d2c828>]






    

In [13]:

    

net.calculate_cost(x_train, y_train)


    

    
Out[13]:





0.28866396260432092

In [14]:

    

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[14]:





[<matplotlib.lines.Line2D at 0x7f73c8d23198>]