Multilayer Perceptron Two-Layer Code Problem

Asked

Viewed 39 times

1

I have a problem with the following code:

import numpy as np

def nonlin(x,deriv=False):
    if(deriv==True):
        return x*(1-x)

    return 1/(1+np.exp(np.float32(-x)))

X = np.array([[488457.495,6673006.568,68.624],[488458.287,6673008.192,68.621],
              [488459.073,6673009.798,68.618], [488456.712,6673004.978,66.558]],dtype=object)# Entrada_Estação_Total_Aba_8_Reg_01

y = np.array([[488457.500,6673006.571,68.624],[488458.281,6673008.199,68.617],
              [488459.071,6673009.807,68.615],[488456.722,6673004.980,66.566]],dtype=object).T#Registro_C1_Nuvem_Aba_*_REG_1

np.random.seed(1)

# randomly initialize our weights with mean 0
syn0 = 2*np.random.random((3,4)) - 1
syn1 = 2*np.random.random((4,1)) - 1

for j in range(60000):

    # Feed forward through layers 0, 1, and 2
    l0 = X
    l1 = nonlin(np.dot(l0,syn0))
    l2 = nonlin(np.dot(l1,syn1))

    # how much did we miss the target value?
    l2_error = y - l2

    if (j% 10000) == 0:
        print ("Error:" + str(np.mean(np.abs(l2_error))))

    # in what direction is the target value?
    # were we really sure? if so, don't change too much.
    l2_delta = l2_error*nonlin(l2,deriv=True)

    # how much did each l1 value contribute to the l2 error (according to the weights)?
    l1_error = l2_delta.dot(syn1.T)

    # in what direction is the target l1?
    # were we really sure? if so, don't change too much.
    l1_delta = l1_error * nonlin(l1,deriv=True)

    syn1 += l1.T.dot(l2_delta)
    syn0 += l0.T.dot(l1_delta)

When I run the code the following error message appears:

Warning (from warnings module):
  File "F:\Perceptron_PE_TLS\Perceptron_TLS_2_Camadas.py", line 7
    return 1/(1+np.exp(np.float32(-x)))
RuntimeWarning: overflow encountered in exp
Traceback (most recent call last):
  File "F:\Perceptron_PE_TLS\Perceptron_TLS_2_Camadas.py", line 39, in <module>
    l2_error = y - l2
ValueError: operands could not be broadcast together with shapes (3,4) (4,1)

When I run the code with integer values this problem does not occur, in case when the values of X and Y are these below the script runs normal.

X = np.array([[0,0,1],
            [0,1,1],
            [1,0,1],
            [1,1,1]])

y = np.array([[0],
            [1],
            [1],
            [0]])
python-3.x networks numpy

1 answer

1


The problem is not whether values are integer or not. The format of y are different.

y = np.array([[0],
            [1],
            [1],
            [0]])
>>> y.shape
(1, 4)

Now with the y of its code (which by the way is equal to X):

y = np.array([[488457.500,6673006.571,68.624],[488458.281,6673008.199,68.617],
              [488459.071,6673009.807,68.615],[488456.722,6673004.980,66.566]],dtype=object).T
>>> y.shape
(3, 4)

Whereas l2 has the format (1,4). The error says exactly this, that it was not possible to do the operation y - l2 because the formats are different.

Browser other questions tagged python-3.x networks numpy

You are not signed in. Login or sign up in order to post.