Continuing with the articles about Neural Networks, this time the Scikit-learn library will be used.

The first step is to install scikit-learn on Linux with pip:

pip3 install  scikit-learn

There will be two examples. The first is to train the neural network with the XOR truth table and see what the network returns; the second exercise takes the data from the previous article (building a neural network in a few minutes) and checks whether the same result is achieved.

XOR will be solved with the K nearest neighbor algorithm (broken link) and also with the multi-layer perceptron Classifier with backpropagation (supervised neural networks).

###A classifier neural network is created to evaluate the truth table for XOR

In [1]:

#Se importa Numpy, MLPCCLassifier y KNeighborsClassifier
import numpy as np
from sklearn.neural_network import MLPClassifier
from sklearn.neighbors import KNeighborsClassifier

In [2]:

#un Arreglo con la tabla de la verdad
# 0 0
# 0 1
# 1 0
# 1 1
xs = np.array([
    0, 0,
    0, 1,
    1, 0,
    1, 1
]).reshape(4, 2)

In [3]:

# Se muestra el arreglo
xs

Out[3]:

array([[0, 0],
       [0, 1],
       [1, 0],
       [1, 1]])

In [4]:

#Se muestra un arreglo conel resultado de hacer un XOR
ys = np.array([0, 1, 1, 0]).reshape(4,)

In [5]:

ys

Out[5]:

array([0, 1, 1, 0])

In [6]:

#Se crea el clasificador con la función de activación relu,con 10k iteraciones y tiene capaz ocultas 4,2
model = MLPClassifier(activation='tanh', max_iter=10000, hidden_layer_sizes=(4,2))

In [7]:

#Se entrena la red neuronal pasando los arreglos de entrada y de salida
model.fit(xs, ys)

Out[7]:

MLPClassifier(activation='tanh', alpha=0.0001, batch_size='auto', beta_1=0.9,
       beta_2=0.999, early_stopping=False, epsilon=1e-08,
       hidden_layer_sizes=(4, 2), learning_rate='constant',
       learning_rate_init=0.001, max_iter=10000, momentum=0.9,
       nesterovs_momentum=True, power_t=0.5, random_state=None,
       shuffle=True, solver='adam', tol=0.0001, validation_fraction=0.1,
       verbose=False, warm_start=False)

In [8]:

print('prediccion:', model.predict(xs)) # salida 0110

prediccion: [0 1 1 0]

In [9]:

print('Se espera:', np.array([0, 1, 1, 0]))

Expected: [0 1 1 0]

####Input and output data from the previous article

Entrada 	Salida
001 	0
111 	1
101 	1
011 	0
100 	?

In [10]:

#Se importa de numpy array
from numpy import array

In [11]:

#Datos de entrada y de salida
datos_entrada = array([[0, 0, 1], [1, 1, 1], [1, 0, 1], [0, 1, 1]]).reshape(4, 3)
datos_salida = array([[0, 1, 1, 0]]).reshape(4, )

In [12]:

#En este caso se usa KNeighborsClassifier con 2 capaz
KNC = KNeighborsClassifier(n_neighbors= 2)

In [13]:

KNC.fit(datos_entrada,datos_salida)

Out[13]:

KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',
           metric_params=None, n_jobs=1, n_neighbors=2, p=2,
           weights='uniform')

In [14]:

#Se predice el valor de 1,0,0 que da como resultado el mismo del artículo anterior.
print(KNC.predict([[1, 0,0]]))

[1]

In [15]:

#Se crea la red de nuevo pero ahora con PLPCCLassifier.
#Se crea el clasificador con la función de activación relu,con 10k iteraciones y tiene capaz ocultas 4,2
KNC = MLPClassifier(activation='tanh', max_iter=10000, hidden_layer_sizes=(4,2))

In [16]:

#Se entrena la red neuronal pasando los arreglos de entrada y de salida
KNC.fit(datos_entrada, datos_salida)

Out[16]:

MLPClassifier(activation='tanh', alpha=0.0001, batch_size='auto', beta_1=0.9,
       beta_2=0.999, early_stopping=False, epsilon=1e-08,
       hidden_layer_sizes=(4, 2), learning_rate='constant',
       learning_rate_init=0.001, max_iter=10000, momentum=0.9,
       nesterovs_momentum=True, power_t=0.5, random_state=None,
       shuffle=True, solver='adam', tol=0.0001, validation_fraction=0.1,
       verbose=False, warm_start=False)

In [17]:

print(KNC.predict([[1, 0,0]]))

[1]

The result is the same using two different types of neural networks; in the case of the multi-layer perceptron the tanh activation function was used.

Make a donation! If you liked the article you can make a donation with Bitcoin (BTC) using the digital wallet of your choice to the following address: 17MtNybhdkA9GV3UNS6BTwPcuhjXoPrSzV

Or by scanning the QR code from the wallet: